Development of a Multichamber Batch Reactor for the Production of Multilayer Interpoly Dielectrics

MCBRIDE - 2403

Keywords multichamber batch reactors, clustered processing, oxynitrides, ONO films, non-volatile memories

Start Date: 15-NOV-88 / Duration: 60 months

[ contact / participants ]

Objectives and Approach

The objective of this project was the development of a multichamber batch reactor in which interpoly silicon oxide and silicon nitride films, combined in a so-called ONO configuration, can be produced in one reactor system.

In state-of-the-art devices, interpoly ONO films are implemented in the floating gate structure of EPROM and EEPROM devices. Following the development of higher circuit integration and the corresponding reduction in dielectrics thickness, it appears that the integrity of the interfaces between the various layers will soon become the limiting factor in the electrical performance (electrical defect density, charge retention) of future devices. It is thus important that the dielectric structure is deposited in a single vacuum system, preventing the wafers from coming into contact with air, before the capacitive structure is completed. In the latter case, in fact, the resulting uncontrolled growth of native oxide films combined with the exposure of the surfaces to chemical contaminants would lead to the degradation of the interface quality in an unreproducible way, adversely affecting the electrical characteristics of the devices.

An R&D prototype was developed in the first phase of the project. It includes five modules grouped around an evacuated central wafer-handling system where wafers are transferred on a single wafer basis from one module to the next. The modules consist of an input/HF cleaner, an oxidation reactor chamber, a low pressure CVD chamber, a staging module and a cassette output.

Progress and Results

The construction of the Mark I prototype and its installation at LETI were accomplished by the end of 1990. Several experiments on process characterisation and hardware evaluation have been carried out at both LETI and ASM. As a result, various hardware upgrades have been performed, leading to a production-quality A600/2 cluster tool that meets stringent industrial requirements. Etching rates and surface concentrations (by XPS and wettability) have been investigated under a wide variety of process conditions to assess process uniformity on the same wafer and from wafer to wafer on complete batches, as well as to evaluate run-to-run reproducibility. Very reproducible etch rates, that are no longer affected by the exact nature of the oxide surface, have been achieved. Etch selectivities for different types of oxides have also been significantly improved. All functional expectations have now been met, and the process on 25 8-inch wafers consistently yields uniformities of less than about 5% wafer-to-wafer, less than about 5% run-to-run, and about 10% within wafer.

In parallel with the optimisation of the HF/H[2]O etch process, ASM has started an investigation of alternative etch chemistries, where water is replaced by other catalytic chemicals. It has been found that the selected alternative chemistry is very attractive, with simple hardware sufficient to produce reproducible etch results. Furthermore, the chemistry is particle and metallic contamination neutral, and polygate structures fabricated using this alternative vapour etch chemistry show excellent E[bd] and Q[bd] characteristics.

Equally important for advanced ONO deposition, where further shrinkage of the dielectric's thickness is required, is the control and optimisation of the nitride film deposition process. For this, the Mark I prototype has demonstrated its excellent capabilities, in that for the first time it has been possible to study the nitride deposition kinetics on clean (ie native oxide free) silicon surfaces. For the ONO application, in particular, it has been concluded that:

Electrical tests showed that the ONO interpoly dielectric fabricated in the A600/2 behaves as well as the ones fabricated in conventional equipment. One significant improvement was found with the shrunk nitride film deposited at 650 °ree;C: even with a very thin (<20 Å) nitride film in the ONO stack, the E[bd]and charge retention values are as good as those obtained with thicker ONO stacks produced in standard equipment. It can therefore be concluded that the nitride process development has enabled a further scaling-down of the ONO stack thickness without affecting the electrical properties. As a result, the capacity per unit area can be increased and thus a higher degree of integration achieved.

A lot of the standard STM A604 test pattern has been used to electrically test ONO dielectrics produced in the clustered tool and to compare this process with the standard process at ST. E[bd], Q[bd] and charge retention characteristics have been evaluated. From the elctrical tests it has been concluded that the Mark I cluster tool is effective in improving the overall quality of the ONO layer compared with the conventional horizontal furnace using the standard flow. Furthermore, the shrunk nitride layer gives superior results. Additional development of the HF etching is necessary, however.

In the last phase of the project, particular attention was given to the development of NO structures (for EPROM/DRAM memory devices) and very thin gate/tunnel oxides, using an in situ HF vapour process based on alternative etched chemistry.


As a result of this project, solutions are expected to some of the technological problems related to the manufacturing of 0.5-0.3 micron non-volatile memory devices (corresponding to 16 and 64 Mbit memories). Furthermore, although this type of reactor has been tested on ONO structures, the proposed system could be used to improve the electrical characteristics of other types of device structures as well (particularly DRAMs). Additional advantages related to the elimination of waiting times in between successive process steps and to the implementation of cassette-to-cassette automation would further increase the industrial potential of this type of equipment.

ASM's new cluster tool, the Advance 600/2, is based on the results of this project and has received a lot of attention from industry: many demonstration runs have been requested by customers interested in the possible advantages of clustered processing. A few systems have already been purchased, notably for the fabrication of NO films and poly contacts and polyemitters. In addition to these, a few more systems are currently being made and soon be installed at customer sites.


Mr L. Kwakman
Rembrandtlaan 2a
tel: + 31/ 30-298523
fax: + 31/ 30-293461
telex: 47374


CEA - F - P

TCS synopses home page TCS acronym index TCS number index
All synopses home page all acronyms index all numbers index

MCBRIDE - 2403, December 1993

please address enquiries to the ESPRIT Information Desk

html version of synopsis by Nick Cook