Measuring Film Thickness using the Opti-Probe

Weilan Ladeau

Optics 464

3/8/03

Abstract:

Without some way to accurately measure the different films we have no control over those factors that determine the conduct of the device. My project is about the Therma-wave Opti-probe. The opti-probe is used too measure thickness of various types of single and double film stacks used in the creation of microchips. The opti-probe looks at the intensity of reflected light and uses the difference in the critical angle to determine thickness. I will discuss theory concerning how the opti-probe uses light to determine thickness and trace the path of the light through the machine.

As I stand at the opti-probe measuring the thickness of my process I wonder how this machine can measure something smaller than on of the hairs on my head. We use it to measure everything from nitride to oxide to polysilicon. Some films are one layer thick and some have multiple layers. Thickness can vary from less than 50 angstroms of oxide to several hundred angstroms of nitride. This project gave me the opportunity to answer the questions I had since starting work at Intel. The optiprobe provided a suitable format for this project.

The Beam Profile Reflectometer, (BPR), measures thickness based on the intensities of reflected light. BPR utilizes the fact that there is s-polarized light and p-polarized light. S-polarized rays are ones in which the plane of incidence is perpendicular to the polarized direction. P-polarized light rays fall along the plane of incidence.

If the angle of incidence is 0 degrees the reflected amplitude for the s and p polarizations are the same. As the angle of incidence increases so does the intensity of the s-polarized rays. For p-polarized waves, however, there is an initial decrease of intensity initially then an increase. As the p-polarized light starts to decrease it reaches a point where the reflected intensity is essentially 0. This angle of incident is called Brewster’s angle and figures prominently in theory of the optiprobe. We obtain the p-polarized rays because of Brewster’s angle. P-polarized rays are refracted through the film and reflected by the substrate because of electrical oscillations introduced in the substrate. Light will not be reflected if it travels in the same directions as the oscillations. These oscillations are such that the vibration is perpendicular to the direction of the ray and parallel to the direction of the reflected ray. Thus, Brewster’s angle (iB)=, the value of this angle provides information on the index n1 of the film.

The Opti-Probe uses the phase difference between the two rays to measure thickness of the film in question. The phase delay depends on the length, 2d, of the extra travel, the velocity of light propagation in the film, and the wavelength of the light. The amplitudes of reflected light can be predicted by using superposition to sum the amplitudes of light reflected from the film interfaces. The thickness is based on summation of the intensities of the two rays, (the square of their amplitudes).

Now we need to find out how the measurement of reflectivity as a function of the angle of incidence is implemented in the Opti-Probe. Let’s look at the BPR diagram.

.

The BPR contains: a diode laser, microscopic lens (80x and 5x) which focuses the light onto the wafer, two array detectors which collect the light reflected from the wafer and two other detectors (one for auto focus control and one that measures the incident light intensity).

The BPR uses a diode laser that emits light at a single wavelength (appprox. 6750A) With the 80x lens; the laser beam is focused onto the wafer over a 0.9 micro- meter diameter spot. Because the 80x lens has large numerical aperture, N.A=0.9,the angle of the come is the largest at 64 degree. (sin(64)=0.9) The beam of light is cylindrical in shape before entering the lens and cone shaped upon leaving the lens. The angles vary from 0 degrees to 64 degrees. Light reflected from the wafer travels through the beam splitter to be collected by the detectors.

The light leaving the lens is a mirror image of the light entering the lens. If the light enters at 45 degrees then it leaves the lens at 45 degrees. It is the same distance from the center of the beam before it enters and after it leaves. In other words the angle of incidence of the ray is related to the distance to the center of the beam.

The light from each angle of incidence is measured by a linear diode arrangement. The array contains 256 diodes spaced 25 microns apart. The center of the diode array measures light at 0 degrees and the diodes on the end measure light reflected at 64 degrees. The light measured from the ray is a curve called a “profile”. A maximum of interference for one angle of incidence shows as a maximum in the profile at the diode corresponding to that angle. This profile is symmetrical.

We have two arrays that collect reflected light. The A-array aligned parallel to the Y direction collects s-polarized light and the B-array aligned parallel to the X direction collects p-polarized light. I have included a diagram of ray alignment before and after reflection. The sample on the left shows the ray before it is focused on the wafer. We can see that the light is in line with the plane of incidence. The sample on the right is the beam after it is reflected from the wafer. In this case the light is perpendicular to the plane of incidence.

Reflected intensity vs. angle of incident is plotted as a profile. The A-array creates an A profile (S-polarized light) and the B-array creates a B profile (B-polarized light). Since the two haves of the profile are essentially mirror images of each other the software in the optiprobe “folds” each in half and takes the average of both. The rays that don’t travel along the XZ or YZ planes are not collected by either array and are ignored. The profiles from each array are not the same but complement each other. With two arrays we get more information than with one array only.

For oxide/silicon there is a maximum in the A-profile. The reflectivity at the maximum is 1. At this point rays 1 and 2 are in phase for that angle and the reflectivity is the same as if there was no oxide layer. The p-profile crosses the reflectivity line 1 for the Brewster’s angle (about 55 degrees).

There is some interference between rays reflected from the two interfaces (air/film and film/substrate). This interference appears as maxima and minima.

If we know the index n1 of the film and we can count the number of cycles in the A profile, we can estimate the thickness of film. or 2kcos (r ) =0,2.

Where K=2(=propagation vector), =laser wavelength 6750A, r=angle of the refracted ray.

I learned an entirely new concept about light angles. I learned what Brewster’s angle is and how we use it to find the thickness of some material. I now know why the film does not reflect P-polarized light. I found out the correlation between the maximum incident angle and numerical aperture. Now when I stand in front of the Opti-Probe I can envision what is happening inside the tool to measure the thickness of my wafers. I know how it uses the reflected intensity of light to give me the thickness of the film I am interested in. I will remember the concepts about Opti-Probe for long time.

Reference:

Opti-probe Level 2 Applications Training Manual, Therma-Wave inc., Fremont, Ca