The Semiconductor Technology Program serves to focus NBS efforts to enhance the performance, interchangeability, and reliability of discrete semiconductor devices and integrated circuits through improvements in measurement technology for use in specifying materials and devices in national and international commerce and for use by industry in controlling device fabrication processes. Its major thrusts are the development of carefully evaluated and well documented test procedures and associated technology and the dissemination of such information to the electronics community. Application of the output by industry will contribute to higher yields, lower cost, and higher reliability of semiconductor devices. The output provides a common basis for the purchase specifications of government agencies which will lead to greater economy in government procurement. In addition, improved measurement technology will provide a basis for controlled improvements in fabrication processes and in essential device characteristics.

ards (ARPA/IC/NBS), addresses critical Defense Department problems in the yield, reliability, and availability of integrated circuits. The DNA-supported portion of th Program emphasizes aspects of the work vi relate to radiation response of electron vices for use in military systems. There considerable overlap between the interests of DNA and ARPA. Measurement oriented activity appropriate to the mission of NBS a critical element in the achievement of objectives of both other agencies.

Essential assistance to the Program is als received from the semiconductor industry through cooperative experiments and techcal exchanges. NBS interacts with indust al users and suppliers of semiconductor & vices through participation in standardi organizations; through direct consultation with device and material suppliers, gover ment agencies, and other users; and throug periodically scheduled symposia and workshops. In addition, progress reports, st as this one, are regularly prepared for suance in the NBS Special Publication 400 sub-series. More detailed reports such as state-of-the-art reviews, literature com lations, and summaries of technical efforts conducted within the Program are issued as these activities are completed. Reports # this type which are published by NBS also appear in the Special Publication 400 subseries. Announcements of availability of all publications in this sub-series are se by the Government Printing Office to those who have requested this service. A reques: form for this purpose may be found at the end of this report.

3.1. Incremental Sheet Resistance

A specimen holder was fabricated to facilitate the measurement of dopant profiles by the incremental sheet resistance method [2]. It was designed so that the repeated sequence of anodic oxidation, oxide stripping with hydrofluoric acid, and sheet resistance measurement can be done without removing the specimen from the holder. A cross sectional view of the holder, which is an adaptation of existing designs [3,4], is shown in figure 1. The critical aspect of the design and fabrication is the achievement of a liquid-tight seal around the area to be anodized. In the apparatus shown the seal is made by pressing the specimen against a TFE fluorocarbon lip (L) by means of pressure applied by a spring (S) in the bottom of the

PROFILES

holder. The holder is intended for use with test structure 30 of Test Pattern NBS-3 (see sec. 6.1.). This structure is a van der Pauw [5] sheet resistor with a diameter of 0.76 mm and symmetrically placed contact arms which extend to contact pads at each corner of the square chip, 2.54 mm (100 mil) on a side. The test structure die (C) is mounted on a TO-5 header (H), which is placed in the holder beneath a 1.25 mm diameter opening in the lip. The 0.4 mm wide seal between the lip and the die protects the contact pads, the bonded lead wires which connect the pads to posts on the header, and the header from the electrolyte used for anodic oxidation.

All external parts of the holder are made of TFE fluorocarbon so that the entire apparatus may be immersed in the electrolyte. Alternatively, the electrolyte acid can be confined to the top reservoir. For removal of the oxide layer, hydrofluoric acid can be poured into the reservoir only. There is a tendency for an air bubble to be trapped just above the specimen when liquids are introduced into the reservoir so it is often necessary to use a small brush to clear away the bubble. Not shown in figure 1 are the wires going to the socket (A) and the plastic tubing and elbows through which the wires run. The holder can be varied for use in either a vertical or horizontal position by appropriate elbow and tubing changes. The apparatus has been tested numerous times with an ethylene glycol mixture as the elec trolyte and found to be adequate. However, much more experience in actual profiling is needed before the design can be considered to be fully satisfactory.

(W. R. Thurber and L. M. Smith'

3.2. Mathematical Models of Dopant Profiles

Work continued in the project to develop a mathematical model which can be solved for

redistribution of boron in silicon durthermal oxidation and diffusion. A comtional scheme described previously (NBS . Publ. 400-8, p. 14) was based on the rical solution of a system of integral tions (NBS Spec. Publ. 400-4, pp. 9-11). y of this scheme has revealed the need .dd several additional quadrature subroues in order to resolve various numerical iculties. Since the program is already ; and complex and hence very difficult to :k, work was shifted to a shorter and è promising finite difference algorithm.

Using this algorithm [6], a computer program has been written for the moving boundary value problem (NBS Spec. Publ. 400-1, pp. 911). In order to solve the two diffusion equations, one in the oxide and one in the silicon, and to satisfy the segregation and conservation of mass boundary conditions, the double sweep method [7] was used. Results of several computer runs are being studied with regard to their convergence to the closed form solution of Grove et al. [8].

(S. R. Kraft* and M. G. Buehler)

NBS Mathematical Analysis Section, Applied

Mathematics Division.

Recall that the experiment is conducted by cooling the junction to a low temperature with a small bias which corresponds to a space-charge width W, then reverse biasing the junction to increase the space-charge

width to W,, W1, and then observing the current as the junction is warmed up [9]. The temperature dependence of the electron density on the defects in the space-charge regions can be found by solving numerically eqs (7a) and (7c) of reference [9]. As an example consider the gold acceptor in silicon for which the electron and hole emission rates, in inverse seconds, are [10]

AND CONTAMINANTS

picted in the figure for two typical cooli rates. Nevertheless, electrons occupy only a very small fraction of the defects and is this portion of the space-charge region, the defects are essentially uncharged. Once cooled to the desired temperature, n. (ED. remains essentially unchanged until the wan up period is begun.

Wb

In the neutral region, essentially all the defects are occupied by electrons since the Fermi energy is well above the defect energy. When the space-charge region is increased from to W, at low temperature the i electron density on the defect centers in this portion, nt, does not change, so the defects are initially essentially fully charged (n, 33

*

ti

Nt).

The initial distribution of charge in both portions of the space-charge region is shove by the dotted line in figure 3. During van up the electron density on the defects in the initially uncharged portion between 0 and Wo, t' follows the appropriate cooldown curve (see fig. 2) until it reaches the value ntf The electron density in the intially charged portion between W and W, "t'

remains essentially constant until it begins to fall rapidly toward the value n at a temperature which depends on the heating rate as shown by the upper curves in