Analysis of interface layer characteristics

Details Analysis of interface layer characteristics Analysis of interface layer characteristics

2001-03-13

2002-10-15

Sherry, Michael (Department: 2829)

Semiconductor device manufacturing: process

With measuring or testing

Optical characteristic sensed

C356S632000

Reexamination Certificate

active

06465265

ABSTRACT:

TECHNICAL FIELD
The subject invention relates to measurements of thin films on semiconductor wafers. More particularly, the invention relates to a new approach for analyzing and characterizing the interface layer between the thin film and the substrate in order to more accurately determine the characteristics of the sample.
BACKGROUND OF THE INVENTION
Various optical metrology devices have been developed for measuring and characterizing thin films on semiconductor wafers. One such tool is described in PCT application WO/9902970, published Jan. 21, 1999. The assignee herein has commercialized the device described in that patent application under the name OPTI-PROBE 5240. This device includes a number of measurement technologies. More specifically, the device includes a beam profile ellipsometer (BPE) (see U.S. Pat. No. 5,181,080); a beam profile reflectometer (BPR) (see U.S. Pat. No. 4,999,014); relatively conventional broad band (BB) and deep ultraviolet (DUV) spectrometers; a proprietary broad band spectroscopic ellipsometer (SE) (see U.S. Pat. No. 5,877,859) and an off-axis narrow band ellipsometer (see U.S. Pat. No. 5,798,837). All of the above-recited patents and PCT applications are incorporated herein by reference.
In order to evaluate a sample based on the measurements taken by the technologies mentioned above, various types of fitting algorithms have been developed. These algorithms start with a model of the expected structure of the test sample. In a simple case, the sample might be a silicon substrate covered with a thin layer of silicon oxide. The algorithm is seeded with information about expected parameters of the two materials such as the thickness, extinction coefficient and index of refraction. Using the Fresnel equations, the algorithm calculates expected measurement values, i.e. what data would be measured by the selected optical technique assuming the original guess of the expected sample parameters was correct. These calculated expected measurements are then compared with the actual measurements obtained with the device. Any differences between the calculated measurements and the actual measurements is an indication that the original guesses of the sample parameters were incorrect. Based on the amount of deviation between the theoretical measurements and the actual measurements, the algorithm will adjust the theoretical parameters of the sample (i.e. change the “guess” of the thickness of the oxide layer) and perform another calculation to determine expected measurement values. This process is repeated in an iterative fashion until the calculated measurement values closely match the actual measured values. At this point, it is assumed that the parameters used to generate the expected measurement values are reasonably close to the actual parameters of the sample. Further details about the use of such fitting algorithms can be found in “Simultaneous Measurement of Six Layers in a Silicon on Insulator Film Stack Using Spectrophotometry and Beam Profile Reflectometry,” Leng et. Al, Journal of Applied Physics, Vol. 81, No. 8, Apr. 15, 1997, page 3570-3578.
In order for this approach to be accurate, it is important that the model selected correspond quite closely to the structure of the actual sample. Each layer in the model is assumed to have certain unique characteristics. In the situation described above, where a thin film oxide layer is deposited on a substrate, it would be acceptable to create a model which accounted for only those two different materials, provided that the thin film layer was relatively thick, i.e. a few hundred angstroms or more. However, this simple modeling approach is not acceptable for much thinner layers. This is because the region between the substrate and the thin film layer defines an interface region which has characteristics somewhat different from either the substrate or the thin film layer. This interface region can be about 5 to 10 angstroms thick and contributes to the sample's response to reflected light. When measuring relatively thin films, on the order of 30 angstroms or less, which is typical thickness of gate oxides used in current state of the art lithography processes, one cannot ignore the presence of the interface layer in creating the theoretical model for analyzing the data.
The need to model the interface layer has been discussed in the past. In the prior art, it was assumed that the interface layer had characteristics similar to the thin film layer. Most researchers treated the interface layer as having essentially the characteristics of the thin film layer, but with a higher refractive index. While this approach has helped to improve accuracy of the modeling over situations where no interface layer is considered, the results have been inadequate for accurately measuring the thinnest of the thin films. In other approaches, the interface layer was treated as having a blend of characteristics from the thin film layer and substrate. The latter approaches did not include representations of the specific electronic structure of the underlying silicon.
Accordingly, it is an object of the subject invention to provide a new modeling approach which significantly improves the analysis of data and provides more accurate measurement results for thin films.
SUMMARY OF THE INVENTION
It has been recognized by the inventors herein that the characteristics of the interface layer should be expanded to include the electronic characteristics of the underlying substrate. Accordingly, when a model is created to analyze samples, particularly in very thin film situations, the interface layer should be characterized as being a combination of the characteristics of the underlying substrate and the thin film.
In the preferred embodiment, applicants believe that the critical point models developed in the past to characterize the interaction between silicon and light are best suited for this approach. As described herein, a five-peak critical point model was used for analyzing test data. The five-peak critical point model is used to help characterize the refractive index and extinction coefficient of the interface layer. It was found that when these modifications were made to the modeling of the interface layer, far more accurate results were achieved in measuring the characteristics of the thin film on the semiconductor.
The invention is carried out as part of the analysis of the data obtained from one or more measurement techniques. More particularly, a theoretical model is set up which includes the substrate, the thin film layer and an interface layer. The interface layer is characterized as having parameters corresponding to both the thin film and the underlying substrate. This model is then used in conjunction with the Fresnel equations to calculate expected measurement data. This theoretical calculated data is then compared to the actual measured data. Differences between the calculated data and the actual measured data are then used to vary the expected characteristics of the sample in an iterative process for determining the actual composition of the sample.
In the examples set forth below, a variety of silicon samples having thermal oxides deposited thereon were measured using a rotating compensator spectroscopic ellipsometer. The results of the new modeling approach demonstrated a high degree of accuracy. It is believed that this approach can be used to analyze a variety of multilayer structures including silicon oxide on silicon as well as other thin films and substrates. It is believed this approach is particularly useful for materials with high extinction coefficients such as hafnium oxide and zirconium oxide. This approach can also be used to model the interface between a dielectric layer and a substrate.
It should be noted that even though this invention relates to modeling an interface layer between a substrate and the adjacent layer, this approach can be applied to multilayer structures. In particular, a model for a three layer sample would include the substrate, the three individual layers and the interface layer betw

Affiliated with

Also associated with

Rating

Analysis of interface layer characteristics does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Analysis of interface layer characteristics, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Analysis of interface layer characteristics will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2921853