U.S. Philips Corporation

Andrew J Walker

Roger Cuppens

Robertus D Verhaar

Alwin N Kronert

The invention relates to a non-volatile memory with floating gate, in particular a Flash-EPROM, in which writing takes place through injection of hot electrons into the floating gate and in which erasing takes place through injection of hot holes. To keep the write and erase voltages sufficiently low, p-type zones which locally increase the background doping concentration of the p-type substrate are provided around the n-type source and drain zones. These p-type zones cause an increased field strength at the drain zone whereby hot electrons are formed at the pinch-off point also at lower voltages. This increased background concentration in addition reduces the breakdown voltage of the pn junction of the source and drain zones, so that hot holes for erasing can be formed by pn breakdown at comparatively low voltages. The device is particularly suitable for being integrated into a signal processing IC manufactured in a standard process, such as a microcontroller.

Semiconductor device having a non-volatile memory and method of

If the process were to be changed for optimizing the injection of hot electrons into the channel of the memory transistor in the manufacture of an integrated circuit having an embedded EPROM, this could lead to degradation of the transistors in the logic, in particular in the case of channel lengths below 1 .mu.m. To prevent this, a process is presented by which the non-volatile memory is optimized without the properties of the logic being affected. For this purpose, a first series of steps is performed in which first the floating gate is defined, followed by the source/drain implantation and a side-wall oxidation for obtaining an oxide spacer on the sides of the floating gate. During these steps, the region of the logic to be formed is uniformly protected against implantation and oxidation by the same poly layer from which the floating gate is made. Then, in a second series of steps, the usual CMOS process is carried out whereby first the gates of the transistors are formed, followed by the necessary source/drain implantations.

Method of manufacturing an integrated circuit having a memory element

Very thin tunnel oxides are used in conventional non-volatile memories to obtain a sufficiently strong tunnelling current to or from the floating gate. Usual thicknesses of the tunnel oxide lie in the 8-10 nm range. The invention renders it possible to use tunnel oxides of a much greater thickness, for example of the order of 20 nm, for comparable tunnelling current values. According to the invention, the tunnelling effect is enhanced by implantation of a heavy, high-energy ion, for example As, into a comparatively thin poly layer of the oxide. During this, Si atoms are propelled from the polylayer into the oxide, so that the oxide is enriched with Si, which causes a major change in the tunnelling characteristics. The same oxide which functions as a gate oxide elsewhere may be used for the tunnel oxide. An important advantage of the invention is that direct contact between the tunnel oxide and photoresist layers necessary during the process is avoided, so that the properties of the tunnel oxide are not or at least substantially not impaired by the photoresist.

Growing community of inventors

Eindhoven, Netherlands

Andrew J Walker