Moritaka Nakamura

Free Radicals in an Inductively Coupled Etching Plasma

A high-density (>1011 cm-3) CF4/H2 plasma was produced in an inductively coupled plasma (ICP) reactor where an external helical coil is wound around a quartz tube. Capacitive coupling from the coil to the plasma caused the release of a large number of impurities ( SiF4 and CO) from the warm quartz wall close to the coil. These impurities significantly deteriorate the etch selectivity of SiO2 to Si in the ICP reactor. Water cooling and a Faraday shield are effective to suppress the release of impurities. Neutral radicals CF3, CF2, CF and F were measured in addition to ionic species. The high-density high electron-temperature ICP causes the formation of a large number of F atoms and CF+ ions with fewer CFx radicals, in comparison to a low-density capacitively coupled plasma (CCP). H2 addition to the CF4 discharge drastically modifies the CF3 and CF2 densities in the ICP as well as in the CCP. The high etch rates and the low selectivity of SiO2 to Si obtained in the ICP were discussed taking account of the residence time and the dissociation time of reactive species in the etching reactor.

Diagnostic of Surface Wave Plasma for Oxide Etching in Comparison with Inductive RF Plasma

Surface wave plasma (SWP) and inductively coupled plasma (ICP) reactors are high plasma density, unmagnetized sources that show promise for use in next-generation etching processes. We compare the 2.45 GHz SWP with the 13.56 MHz ICP in terms of the radical composition in C4F8/Ar discharges and the electron energy distribution function (EEDF). A comparison of the two plasmas was carefully made in an identical plasma vessel at the same wall temperature where an antenna coupler on a quartz plate was changed from an expanded waveguide for SWP to a loop coil for ICP. Reactive species measurement at the same electron density under the same gas conditions showed marked differences. First, the dissociation degree of C4F8 at the same electron density is higher in ICP than in SWP. Second, neutral radical densities (CF3, CF2) at the same electron density are several times higher in SWP than in ICP, and ICP has a high F radical density. Third, as regards ionic composition, ICP contains more Ar+ and less fluorocarbon ions (CmFn+), while large molecular ions (C2F4+, C3F3+, C3F5+) exist in SWP. In conclusion, ICP is more dissociative than SWP at the same electron density. This result is tentatively attributed to the difference in the EEDFs of the two plasmas, since optical emission spectroscopy of Ar I suggests 1.5–2 times more high-energy (>14 eV) electrons in ICP than in SWP.

Resist stripping in an O2+H2O plasma downstream

Characteristics are reported for a resist stripping process downstream of an oxygen plasma to which water vapor is added. The effects of additive water vapor are an increase in atomic oxygen concentration in the plasma, a decrease in activation energy of ashing reaction, and protection of semiconductor devices from the sodium contamination from the resist. The atomic oxygen concentration was approximately doubled by mixing 10% H2O into the oxygen plasma. The activation energy of the ashing reaction to the resist made from novolak resin decreased from about 0.5 to 0.39 eV by the addition of water vapor of more than 1%. The activation energy of hydrogen abstraction from hydrocarbon molecules by an OH radical was lower than that by a ground state oxygen atom [O(3P)], which was the dominant ashing species in the oxygen plasma downstream, and that by an atomic hydrogen was higher than that by the ground state oxygen atom. Moreover, the activation energy in the downstream ashing of the oxygen plasma added to which was 1% water vapor was lower than that of the oxygen plasma to which 3% hydrogen was added, even though the relative concentration of atomic hydrogen in each plasma was equal. Therefore the decrease in the activation energy was probably due to the OH radical generated in the plasma and the downstream. Sodium atoms in the resist were blocked from entering into the semiconductor devices in the stripping process by use of the O2+H2O plasma downstream. Thus sodium was not removed and remained on the wafer surface after resist stripping. Also, by adding N2 or CF4 to the O2+H2O plasma, we can increase the ashing rate without losing the above characteristics.

Additive Nitrogen Effects on Oxygen Plasma Downstream Ashing

Using an improved actinometry method, additive nitrogen effects on oxygen plasma downstream ashing have been studied. The ion current of the Langmuir probe and emission intensity change in OI(7774) and OI(8446) as a function of nitrogen mixing ratio showed that emission caused by dissociative excitation of oxygen molecules did not significantly influence the actinometry in our experiment. Thus, the actinometry measured accurate relative concentrations of atomic oxygen in the plasma by selecting XeI(4671) or XeI(4624) for the actinometer to OI(7774) and OI(8446) or by using ArI(7503) or ArI(7067) for the actinometer to OI(6258) and OI(4368). The change in the ashing rate and the relative concentration of atomic oxygen as a function of the nitrogen mixing ratio corresponded well, and both values at 10% nitrogen mixing were twice those with no nitrogen mixing. The activation energy was unchanged regardless of additive nitrogen. Therefore the role of nitrogen as the additive impurity gas is only to increase oxygen in the plasma.

Direction of topography dependent damage current during plasma etching

We have studied the electron shading or topographical effects on charging damage by measuring the damage current directly using newly developed probes set on the wafer surface. By using the probe to model the end point of metal etching with high-aspect-ratio patterns, positive current flowing from the antenna to the substrate was observed. On the contrary, with the second probe for overetching of low-aspect-ratio patterns, negative current flowed from the antenna to the substrate. The negative damage current and charging voltage for the second probe increased with increasing space width and electron temperature. These results explain the recently reported complex phenomena in device damage measurement and give more insight into electron shading effects.

Very High Selective n+ poly-Si RIE with Carbon Elimination

The effect of carbon contamination was studied in low-temperature HBr RIE of n+-doped poly-Si etching. With a resist mask, which is a notable carbon source, selectivity (poly-Si/SiO2) was 10 to 20. With a SiO2 mask and carbon elimination from various sources such as gas, piping system and reactor wall, SiO2 etching rate was suppressed, and selectivity of more than 300 was achieved with a self-bias voltage of 400 V. Addition of a small amount of oxygen after carbon elimination removed residual carbon and enhanced the selectivity up to 3000. The anisotropic profile was not affected by carbon elimination since the side-wall protection by reaction products (SiBrx ) was the mechanism of anisotropy. The effect of carbon on the selectivity was explained by the thermodynamics on the basis of bond strengths of reaction products.

Reduction of Electron Shading Damage Using Synchronous Bias in Pulsed Plasma

A novel method is proposed for reducing charging damage due to the “electron shading” effect. The concept is to selectively utilize the coolest electrons in a pulsed plasma at the end of its off period by synchronizing rf bias; thereby one should be able to reduce the negative charge build-up responsible for the damage. This concept has been examined using an inductively coupled plasma (ICP) apparatus. Exposure to a cw Ar ICP damaged most MOS capacitors with a 6-nm-thick gate oxide and connected to 105 shaded antennas. This damage was reduced only slightly even with 5-µ s- on/10-µ s- off pulse modulation when the rf bias was asynchronous (60 kHz). When the rf bias (66.7 kHz) synchronized at the expected optimal phase, the most significant reduction of the damage was observed. This effect is discussed based on the results of time-resolved probe and optical emission measurements.

Chemical States of Bromine Atoms on SiO2 Surface after HBr Reactive Ion Etching: Analysis of Thin Oxide

Low-energy ion scattering spectroscopy (ISS) and X-ray photoelectron spectroscopy (XPS) have been used to determine the nature of Br atoms on very thin thermal silicon dioxide (approximately 5 nm) after HBr reactive ion etching (RIE). The result of ISS clarified that the etched surface was covered with 1 monolayer Br. The Br atoms on the etched SiO2 surface were found, from the result of XPS analysis, to have two chemical adsorption states. The experiment of atomic Br exposure showed that one adsorption state was on the as-grown SiO2 surface and the other state was at the damaged sites induced by ion bombardment.

FRAM technologies compatible with 0.5-um CMOS logics

We developed FRAM (FRAM is a registered trademark of Ramtron International Corporation that stands for FeRAM) technologies that are fully compatible with half-micron CMOS logics. The technologies achieve 1T/1C FRAM cell 12.5 micrometer2 in a size and 68k-FRAM embedded 8bit-MCU. The CMOS transistors work at 5V for a cell operation and 3V for a logic operation. We did not use a COB to employ a present CMOS processing, and used the local interconnect to reduce a chip size. We used the W plug to contact to deep diffusion layers through high-aspect contact holes. The CMP planarization was used to relax PZT deposition and Pt etching. To prevent the process degradation of PZT, we used single Al wiring with SOG as an interlayer dielectric. The cover dielectric was formed with plasma TEOS- CVD without SiN to prevent the process degradation at this case. The SiN cover will be indispensable in real products. These technologies achieved a cell size 6.95 X 1.8 equals 12.5 (micrometer2) for 1T/1C and 4.2 X 6.5 equals 27.3(micrometer2) for 2T/2C that are the smallest cell size in FRAMs that do not use a COB structure and a poly-plug as a storage.

Sodium contamination free ashing process using O2+H2O plasma downstream

The effect of O2+H2O plasma downstream ashing on sodium contamination from resists has been studied through measurement of flatband voltage shifts of metal–oxide–semiconductor (MOS) diodes after bias‐stress treatment and measurement by atomic absorption spectroscopy of the amount of sodium in the SiO2 layer after resist stripping. The flatband voltage shift of the MOS diodes on which a resist layer was stripped by the O2+H2O plasma downstream was almost the same as that treated by the O2+H2O plasma downstream without resist and smaller than that where the resist was removed by other dry ashing methods such as O2 plasma and O2 plasma downstream. In addition, the amount of sodium that existed in the SiO2 layer after resist ashing by the O2+H2O plasma downstream was also nearly the same as that in the SiO2 layer as grown. This sodium passivation by the O2+H2O plasma downstream was most effective at the H2O percentage of 40%–60%, and did not depend on wafer temperature below 200 °C or over‐ashing time. These ...