AVS2007 Session PS-MoA: Plasma Processing for High k, III-V and Smart Materials

Monday, October 15, 2007 2:00 PM in Room 607

Monday Afternoon

Start	Invited?	Item
2:00 PM		PS-MoA-1 Activation Energies for HfO₂ and Si Etching in BCl₃ Plasmas, and Boron Cleaning from Si in H₂ Plasmas C. Wang, V.M. Donnelly (University of Houston) We have investigated plasma etching of a high dielectric constant material, HfO₂, as well as poly-Si in BCl₃ plasmas. Etching rates of HfO₂ and poly-Si were studied as a function of substrate temperature (T_s) and plasma source power, and activation energies for HfO₂ and poly-Si etching were measured at several powers. There is only a slight increase in the etching rate of HfO₂ and poly-Si with increasing temperature. Activation energies range from 0.2 to 0.9 kcal/mole for HfO₂ and 0.8 to 1.8 kcal/mole for Si, with no obvious dependence on source powers over the range studied (20 to 200 W). These low activation energies suggest an etching mechanism in which product removal is limited by chemical sputtering of the chemisorbed layer on the surface and higher T_s modestly increases the reaction rate during the ion “thermal” spikes. H₂ plasma cleaning of the thin B-containing layer remaining after BCl₃ plasma etching of HfO₂ on Si was also studied. Previously, we have reported that B can be cleaned from Si in dilute H₂/Ar (1% H₂) plasmas in 20 s at room temperature, provided the reactor chamber was cleaned in pure H₂ plasmas first with sample absent. Here we present a study of boron cleaning in dilute H₂/Ar plasmas at elevated substrate temperatures, using x-ray photoelectron spectroscopy to measure B removal rates for individual BCl_xO_y moities. We have found that the B cleaning rate is faster at higher T_s. An activation energy of 2.7 kcal/mole was obtained for total B removal in a 1% H₂/Ar plasma. Conversely, the Si etching rate under these conditions displayed little if any dependence on substrate temperature; the activation energy was between 0.2 and -0.6 kcal/mole. Therefore, it is advantageous to remove B at higher T_s to minimize Si removal. For example, at T_s = 235 °C, ~90% of B is cleaned from Si in less than 10 s, while <10 Å of Si is removed in this period. Moreover, it was found that etching of Si stops and a Si-oxide layer forms if oxygen is present in the H₂ plasma (e.g. from erosion of silica components in the reactor). Consequently, still higher selectivities of B removal with respect to Si are possible under conditions where a small amount of oxygen is present in the H₂/Ar plasma.
2:20 PM		PS-MoA-2 Analyses of Deposition/Etching Regimes during Selective Etching of HfO₂ on Silicon in BCl₃ Plasmas: Impact of Chamber Walls E. Sungauer (ST Microelectronics, France); X. Mellhaoui, E. Pargon (LTM/CNRS, France); Th. Lill (Applied Materials Inc.); O. Joubert (LTM/CNRS, France) With the continuous scaling down of CMOS devices to ensure higher speed and density, the thickness of the SiO₂ gate dielectric is expected to be reduced down to 1nm for the 45 and 32nm technological nodes. This thickness reduction brings some serious issues such as increased gate leakage current and reduced oxide reliability. Therefore, high-k metal oxides, and more particularly HfO₂ have been considered as alternative materials to provide substantially thicker dielectric layers for reduced leakage current and increased gate capacitance. The present work focuses on the understanding of HfO₂, SiO₂ and Si etching mechanisms in BCl₃ based plasmas. BCl₃ seems to be a promising gas providing high etch selectivity between HfO₂ and Si substrates. The 200mm wafers are etched in an industrial ICP reactor, and then transferred under vacuum into an X-ray Photoelectron Spectroscopy (XPS) analysis chamber to investigate surface modifications induced by plasma exposure. XPS experiments help us in understanding the mechanism driving the etch selectivity between HfO₂ and Si-containing substrates. The role of Boron is fundamental since Boron by reacting with Silicon and forming Si-B bonds favour the growth of BCl_x polymer on Silicon surfaces slowing down Silicon etching. On the other hand, on HfO₂ surfaces Boron is directly involved in the etching by helping the formation of volatile BOCl etch products. The ionic bombardment plays also a key role since it controls the BCl deposition rate. The ion energy threshold which controls the transition between etching and deposition is lower on HfO₂ than on Si and SiO₂ wafers, implying that infinite etch selectivity between HfO₂ and Silicon can be obtained if the ion energy is well adjusted. In-situ kinetic ellipsometric measurements were also carried out on HfO₂, SiO₂ and Si substrates to monitor in real time the etching/deposition transition during BCl₃ plasma exposure. These experiments have revealed that the etch or deposition rate is linear with time only after a transient regime of about 10s and that during the10 first seconds, HfO₂, Silicon and SiO₂ show very different kinetic behaviors. We also observed that reactor wall conditioning plays a key role in controlling BCl_x deposition on the wafer and that infinite selectivity can be obtained by coating the reactor walls with carbon layer prior etching in BCl₃.
2:40 PM		PS-MoA-3 Optical Emission Study of an Inductively Coupled Cl₂/H₂ Plasma during InP Etching of Micro-nanostructures used for Photonic Applications L. Gatilova, S. Bouchoule, S. Guilet (Laboratoire de Photonique et de Nanostructures (LPN)-CNRS, France); P. Chabert (Laboratoire de Physique et de Technologie de Plasmas (LPTP)-CNRS, France) Cl2/H2-based chemistry has proven to be very efficient for highly anisotropic ICP etching of InP-based heterostructures used in photonic devices. It was shown recently that the Cl2/H2 ratio is a key parameter to control the sidewall profile. At low pressure (0.5mT-1mT), the onset of anisotropic regime occurs at H2 = 35-45%, where the evolution of the etch rate with H2 percentage shows a maximum. A possible explanation, proposed in literature, is the decrease of the reactive atoms and ions (Cl, Cl+) because of the by-products (i.e. HCl) formation. However, deeper understanding of InP etching mechanism requires more detailed investigations. We have used OES combined with electron and positive ion density measurements, during the etching of InP ridge structures, to obtain insight into the etch mechanism of InP in Cl2/H2 ICP plasma. The pressure was 0.5 mT, the ICP power was 800W, the DC bias voltage was –150 V, the total gas flow was kept constant at 28 sccm, and the H2 concentration is varied from 0 to 100%. The main emission lines recorded during the etching process were Cl (725.7 and 754.7nm), H (656.3nm), In (325.6, 410.2, 451.1nm), InCl (350nm), PH (340nm). In order to estimate the relative atom concentrations, 10% of argon was added in the initial gas mixture. The etch rate and the In-line intensity have roughly the same behavior versus Cl2/H2 ratio, which can be divided into three regions. For H2 concentration between 0-25% (corresponding to strongly undercut profiles), the etch rate rapidly decreases with %H2 increase, so as the positive ion current and the reactive species concentration – the Cl density falls down continuously when H2 increase from 0 to 100%. For H2 concentration greater than 60%, the etch rate also decreases down to very low values < 100nm/min and the etched surface becomes grassy. Despite the H2 concentration increases, the concentration of H atoms decreases, probably due to the decrease in the electron density. For intermediate H2 concentration (the second region which lies between 35% and 45%), corresponding to the highly anisotropic region, the etch rate remains constant. This intermediate region corresponds to a maximum in H concentration. The etch rate could thus be the result of a balanced effect between the Cl density decrease and the H density increase, with a change in etching mechanisms of P-atoms; for high H density P-atoms leave the InP surface by PHx formation, as suggested by the increase of PH-line intensity.
3:00 PM		PS-MoA-4 Dry Etching of Ge₂Sb₂Te₅ for Phase Change Memory Applications: Characterization and Design of Low Damage Process P. Petruzza (STMicroelectronics Italy) In order to realize highly integrated PRAM involving Ge₂Sb₂Te₅ (GST) thin films,footnote1 the etching process must be developed. Until now, there were several work devoted to the investigations of etching properties of GST films using fluorine and chlorine based plasma chemistries.footnote2 Unfortunately, the relationships between plasma parameters and damage of GST thin films remained out of attention. We investigated the etching behaviours of GST in terms of etching process parameters such as pressure, gas, temperature, gas flux directionality in closely bound up with film stack of GST and subsequent problems such as voiding, poisoning of GST ad decreased mechanical strength. Etching of chalcogenide alloy may result in chemical and structural modification of the sidewall and surface residues. GST reactive ion etching plasmas have been studied by measuring etch rate and composition using XRF spectroscopy, etch profile, surface - chemical aspects and bulk morphology by employing TEM/SEM. Etching experiments were performed in a low pressure inductively coupled plasma reactor supplied with 13.56 MHz rf powers. GST thin films were prepared on SIN substrate. Damage and degradation of GST has been investigated by down stream plasma treatments after GST etching definition. SEM cross section analysis shows that the sidewalls of GST are eroded after ashing process. The thickness of damaged thin layer depends of etching chemistry and of other process parameters such as temperature. In the present work it’s explained the results obtained with this method for different etching gas chemistry of chalcogenide alloy. The experiments results show that Cl₂ etching process have induced a composition change of the alloy and thick erosion in sidewalls of GST film patterns. Unlike Cl₂, fluorine chemistry avoids GST erosion. In order to perform manufacturability phase change memory: using the obtained results, a etching process by Cl₂ free chemistry with suitable process parameters has been provided. footnote1 F. Pellizer, A Pirovano, et all., Proc. Symposiumon VSI technology, june 2004, pg 18-19. footnote2 Sung-Min Yoon et al., Japanese Journal of Applied Phisics, Vol. 44, No 27, 2005, pp L 869-L 872 .
3:40 PM	Invited	PS-MoA-6 John A. Thornton Memorial Award Lecture - Etching of SiC, GaN and ZnO for Wide Bandgap Semiconductor Device Applications S.J. Pearton, L.F. Voss, W.T. Lim (University of Florida); R.J. Shul (Sandia National Laboratories) A review will be given of dry etching of three technologically important wide bandgap semiconductors, namely GaN,SiC and ZnO. Dry etching of GaN is needed for mesa formation on electronic and photonic devices and for through-wafer vias on power devices. Generally chlorine-based plasma chemistries are used, with etch rates in the range of a few thousand angstroms per minute to almost one micron per minute. A typical issue is the preferential loss of nitrogen from the near-surface region, leading to the presence of an n-type surface layer after etching. This can be used to advantage in improving contact resistance of n-type Ohmic contacts. For SiC, the main chemistries are based on fluorine and changes to the surface electrical properties are less of an issue. For ZnO, the low volatility of all Zn etch products leads to low etch rates at room temperature and changing to iodine or bromine chemistries does not improve the removal rates. Examples will be given of device etching processes for all three materials systems.
4:20 PM		PS-MoA-8 Comparative Study of ECR and ICP Plasma Etching of High-k Dielectric HfO₂ Films with BCl₃-Containing Gas Chemistries D. Hamada, K. Nakamura, Y. Ueda, M. Yoshida, K. Eriguchi, K. Ono (Kyoto University, Japan) Etching of high-k materials is indispensable for their removal in integrating them into device fabrication. Moreover, the high-k etching is required for chamber cleaning of the deposition apparatuses in mass production. This paper presents a comparative study of the etching of high-k HfO₂ films in electron cyclotron resonance (ECR) plasma and inductively coupled plasma (ICP) reactors with BCl₃-containing chemistries, where emphasis is placed on a better understanding on the etching mechanisms concerned. The ECR reactor had a configuration of divergent magnetic fields, and the discharge was established by 2.45-GHz microwave powers of 600 W. The ICP reactor had a three-turn planar coil, and the discharge was established by 13.56 MHz rf powers of 300 W. Feedstock gases were BCl₃, Cl₂, O₂, and Ar at total pressures of 2-20 mTorr with a total flow rate of 40 sccm. The significant differences between ECR and ICP plasmas are: the etching of HfO₂ without rf biasing was obtained in ECR BCl₃-containing plasmas, while was not obtained in ICP; moreover, the etch selectivity HfO₂/Si was >> 1 with no bias in ECR, while was < 1 with bias in ICP. In ECR, the HfO₂ etch rate was increased in order of BCl₃, BCl₃/O₂, BCl₃/Cl₂, and BCl₃/Cl₂/O₂; typically, the HfO₂ etch rate in BCl₃/Cl₂ was ~100 nm/min at ~60% Cl₂ with a selectivity of ~10 over Si, and a high selectivity >50 was obtained at 40-50% Cl₂ with a HfO₂ etch rate of ~50 nm/min. The Langmuir probe measurements indicated that in ECR, the difference between the plasma and floating potentials was of the order of 10 V, which is lower than the threshold ion energy ~26 eV known for the HfO₂ etching in BCl₃ plasmas. In contrast, the HfO₂ etching in ICP occurred with additional rf biasing, where the threshold energy was estimated to be ~ 30 eV from the difference between the plasma potential and dc self-bias voltage; the etch rate increased with increasing rf bias power, being ~50 nm/min with a HfO₂/Si selectivity of ~0.5 at an ion energy of ~100 eV in BCl₃/Cl₂. The gas-phase and surface chemistries responsible for the HfO₂ etching is discussed based on several plasmas and surface diagnostics including OES, QMS, LIF, FTIR, and XPS, to achieve higher etch rate and selectivity under conditions of low ion energies and/or less ions.
4:40 PM		PS-MoA-9 Investigation of Surface Reactions for Chlorine-Based Plasma Etching of Nitrided Hafnium Silicates R.M. Martin (University of California at Los Angeles); B. Xia, A. Misra (Air Liquide); J.P. Chang (University of California at Los Angeles) The development of plasma etching chemistries is necessary to pattern new gate dielectric materials, such as hafnium-based oxides, for sub-45nm CMOS devices. Nitrided hafnium silicates (HfSiON) are promising since they combine the high dielectric constant and improved interface state density of hafnium silicates with the beneficial properties of silicon oxynitrides. In this work, chlorine-based chemistries are used in an electron cyclotron resonance high density plasma reactor to etch Hf-rich and Si-rich nitrided hafnium silicates, with 0 to 15 at.% of nitrogen. The plasma density, electron temperature, and gas phase species are characterized by a Langmuir probe, optical emission spectroscopy, and quadrupole mass spectrometry. The etching of SiO₂ and HfO₂ was first studied in Cl₂ and BCl₃ plasmas, to allow for studies of the etching of HfSiON with well controlled and varying compositions of Si and N in HfO₂. The etch rates of nitrided hafnium silicates were found to increase with the square root of ion energy, and the etching rate of films increased with increasing nitrogen incorporation as well as SiO₂ percentage in the film. The surface chlorination was enhanced with increasing ion energy, ranging from 1 to 4 at.% of chlorine on the etched surfaces, demonstrating that the etching reaction is limited by the momentum transfer from the ions to the film surface. The measured etching threshold energies were higher than that of pure HfO₂, suggesting that Si and N incorporation modifies film structure/density. In addition, nitrogen was detected removed in the form of SiN₂Cl_x, and more nitrogen remains on the surface of the Hf-rich films than the Si-rich films. This suggests that the removal of N is related to its bonding within the film. Hafnium and silicon were removed as HfCl_x, SiCl_x, and SiO₂Cl_x, and increased with ion energy. A generalized phenomenological model will be presented to describe the effect of SiO₂ and N incorporation on the etching behavior of HfO₂.
5:00 PM		PS-MoA-10 Plasma Source-Dependent Charging Damage Polarities in the Performance Degradation of MOSFETs with Hf-based High-k Gate Dielectrics M. Kamei, K. Eriguchi, H. Fukumoto, K. Ono (Kyoto University, Japan) We report that the polarities of charging damage in n- and p-ch MOSFETs with Hf-based high-k gate stack (HfAlOx/SiO₂) depends on plasma sources, in contrast to those with conventional SiO₂. In order to investigate the charging polarity in MOSFETs with the high-k gate stack (high-k) and those with SiO₂ (SiO₂), the gate leakage current, drain current - gate voltage, and capacitance - voltage measurements were conducted for at least 12 difference devices with different device sizes (antenna ratio) to evaluate the deviation. The electrical thicknesses by capacitance-voltage measurements are ~2.7 and ~7.4 nm for high-k and SiO₂, respectively, while both devices have approximately the same physical thickness of 7 nm. ECR with the bias power of 200 W under two plasma conditions, Ar- and Cl-based gas mixtures, were utilized to induce the charging damage. The Langmuir probe and bias voltage measurements were carried out for correlating the electrical data to plasma parameters to understand the mechanisms. For Ar-plasma, high-k gate stacks were identified to suffer from negative charge trapping for both n- and p-MOSFETs, while SiO₂, from positive charge trapping for pMOSFET. For Cl-plasma on the other, positive charge trapping was observed for n- and p-MOSFETs with high-k, in contrast to Ar-plasma. The observed unique features in high-k were attributed to the difference in the measured ion currents and electron densities between Ar and Cl plasmas as well as the effects of device polarities (n-/p-ch) and asymmetric energy band structures of high-k gate stacks on the stress configurations during plasma exposures. It is suggested that Ar- and Cl-plasmas exhibit different current sources (positive, negative, or bi-directional current sources) in response to device structures, in particular with high-k subject to charge trapping. In addition to the experimental result that high-k devices are more susceptible to plasma charging damage compared to SiO₂ devices, it can be concluded that the observed plasma source-dependent charging polarity for high-k devices, in particular pMOS, should be considered in future device design rules and plasma process designs.