AVS1997 Session EM+SS-TuM: Role of Hydrogen at Semiconductor Surfaces and Interfaces

Tuesday, October 21, 1997 8:20 AM in Room C3/4

Tuesday Morning

Start	Invited?	Item
8:20 AM		EM+SS-TuM-1 The Structure of Hydrogen-Passivated High-Index Silicon Surfaces A. Laracuente, S.C. Erwin, L.J. Whitman (Naval Research Laboratory) Si(114) was recently found to have a planar (2x1) reconstruction composed of alternately "rebonded" and non-rebonded double-layer "B-type" steps, with a row of dimers between each step.¹ The rebonded steps incorporate an extra row of atoms that reduce the dangling bond density but cause tensile surface stress; the non-rebonded steps allow stress relief at the expense of additional dangling bonds. As a first step in investigating the potential of high-index Si surfaces for electronic device applications, we have used STM and first-principles theoretical methods to study their structure following exposure to atomic hydrogen. On Si(114), exposure at ~200 °C appears to create a mixture of monohydride and dihydride phases. Between 300 °C and 400 °C we observe the disappearance of the dihydride which etches away some of the dimers. At ~450 °C, we can easily prepare a well-ordered, low defect density, monohydride Si(114):H-(2x1) surface composed of H-terminated dimers and H-terminated non-rebonded steps; by passivating the bonds at the step edge, H adsorption eliminates the driving force for step rebonding. The removal of rebonded steps appears to be a general phenomenon, also occurring upon H adsorption on Si(115) and Si(117), and may allow the planarization of otherwise unstable (faceted) high-index Si surfaces. Given recent results showing that thick Ge films grown on Si(118) have novel electronic properties,² our results indicate that high-index Si surfaces may actually make interesting substrates for electronics. ¹S. C. Erwin, A. A. Baski, L. J. Whitman, Phys. Rev. Lett. 77, 687 (1996). ²Thorton, et al., Appl. Phys. Lett. 70, 1278 (1997).
8:40 AM		EM+SS-TuM-2 Deuterium Diffusion in Annealed NMOS Devices: A SIMS Study G.C. Abeln (University of Illinois, Urbana-Champaign); I.C. Kizilyalli, F.A. Stevie (Bell Laboratories, Lucent Technologies); K. Hess, J.W. Lyding (University of Illinois, Urbana-Champaign) The substitution of deuterium for hydrogen in the standard post-metallization annealing process has been shown to produce dramatic lifetime improvements for NMOS transistors¹^,². The purpose of this paper is twofold. First, unpublished channel hot carrier stress experiment results are presented from a third set of deuterium annealed wafers to further confirm the giant isotope effect reported earlier. Second, the mechanism, not yet fully understood, for deuterium transport from the wafer surface to the Si/SiO₂ interface will be elucidated. Secondary ion mass spectrometry (SIMS) was used to study the diffusion of deuterium through multilayer device structures. For typical annealing temperatures of 400-450 C, deuterium accumulates either at the interface between the first interlevel dielectric and the polysilicon gate, or at grain boundaries inside the polysilicon. Higher temperatures are required to incorporate appreciable amounts of deuterium at the Si/SiO₂ interface. These observations suggest that deuterium reaches the gate oxide/ substrate interface along another path, perhaps by diffusing through the sidewall (TEOS) spacers. Ion implantation of deuterium is proposed and demonstrated as an alternative means to introduce deuterium at the Si/SiO₂ interface. This process has the advantage that diffusion barriersin the structure are avoided. ¹J. W. Lyding, K. Hess, and I.C. Kizilyalli, Appl. Phys. Lett. 68, 2526 (1996). ²I. C. Kizilyalli, J. W. Lyding, and K. Hess, IEEE Electron Device Lett. 18, 81 (1997).
9:00 AM		EM+SS-TuM-3 SIMS Characterization of the Deuterium Sintering Process for the Enhanced Life-time CMOS Transistors J. Lee (University of Illinois, Urbana); S. Aur, R. Eklund (Texas Instruments); K. Hess, J.W. Lyding (University of Illinois, Urbana) We have demonstrated the effectiveness of deuterium(²H) sintering in enhancing the hot carrier reliability of NMOS transistors with oxide sidewalls. Through the post-metal sintering in a ²H environment instead of the usual hydrogen passivation, the ²H diffuses rapidly through the interlevel oxides and replaces the existing hydrogen(¹H) at the gate SiO₂/Si interface. This results in a strong kinetic isotope effect which enhances device lifetimes by an order of magnitude or more. As Si₃N₄ is believed to be a barrier in the diffusion of ¹H as well as ²H, the improvement in the lifetime of CMOS transistors with nitride sidewalls is not as apparent as the devices with oxide sidewalls. ²H must overcome this diffusion barrier to reach the SiO₂/Si interface. Otherwise, the passivation of the interface through ¹H that is present in large quantities in plasma deposited dielectric films will ensue. More rigorous sintering parameters are required to overcome the threshold ratio of ²H/¹H concentrations at the interface, as well as a more comprehensive characterization of the sintering process. In our experiment, sintering temperatures varied from 400°C - 480°C which is close to but lower than the temperature limit of post-metalization. The ²H concentration inside the furnace ranged from 10% in ultra-high-purity nitrogen to 100%. The sintering time also differed from 30 - 150 min. We have documented the ²H incorporation at the gate SiO₂/Si interface of CMOS transistors with nitride sidewalls and have found the mean-lifetime improvement in these devices by a factor of 10. Secondary ion mass spectrometry (SIMS) depth profiles of ¹H and ²H were performed by applying a Cs+ primary ion beam and the negative secondary ions in a CAMECA IMS-5f system. In this paper, we will present the lifetime improvements of the devices as a function of the sintering parameters and demonstrate that these results correlate well with the ratio of ²H/¹H concentrations at the interface, as determined by SIMS characterization.
9:20 AM	Invited	EM+SS-TuM-4 Hydrogen in Silicon: Fundamental Properties and Consequences for Devices C.G. Van de Walle (Xerox Palo Alto Research Center) It was recently demonstrated that incorporation of deuterium (D), rather than hydrogen (H), at the Si/SiO2 interface leads to significant improvements in the lifetime of metal-oxide-semiconductor (MOS) transistors¹. Scanning-tunneling-microscope (STM) desorption of H or D from Si(100) surfaces exhibits a similarly large isotope effect. These observations have generated a wave of new interest and activity in the role of hydrogen in passivating surface and interface states. The hydrogen/deuterium issue poses some profound physics questions, in particular: what is the mechanism behind such a giant isotope effect? A brief introduction will cover the fundamental properties of hydrogen interstitials and their interactions with defects and impurities, as well as the information that can be extracted from state-of-the-art first-principles calculations. Then I will discuss the vibrational properties, as well as formation and dissociation of Si-H bonds. An investigation of the dissocation path turns out to provide a natural explanation for the isotope effect². Connections between the phenomena at surfaces, interfaces, and in amorphous materials will be pointed out. I will also discuss how investigations of hydrogen interactions with Ge can shed light on the basic mechanisms. ¹J. W. Lyding, K. Hess, and I. C. Kizilyalli, Appl. Phys. Lett. 68, 2526 (1996). ²C. G. Van de Walle and W. B. Jackson, Appl. Phys. Lett. 69, 2441 (1996).
10:00 AM		EM+SS-TuM-6 Hydrogen Desorption from the H-terminated Si(001)-3x1 Surface T.-C. Shen (University of Illinois, Urbana); Ph. Avouris (IBM Research Division) The H-terminated Si(001)-3x1 surface consists of alternating rows of monohydride and dihydride units. It provides an opportunity to directly compare the H-desorption processes from these two different Si-H bonding states. Using the scanning tunneling microscope (STM) to induce electron stimulated desorption, we can selectively remove H-atoms with atomic scale resolution. We find that the desorption yield of H from monohydride units is a few times higher (<6) than that from the dihydride ones. This is in contrast to the case of thermal H₂ desorption where desorption from dihydride sites takes place at ~100K lower temperature than that from monohydride. When all the H is removed from the 3x1 units, the underlying Si-Si σ-bonds rearrange at room temperature to form bare Si-dimers in a 2x1 structure. To investigate the desorption pathways involving the dihydride on the 3x1 surface, we used the STM to remove two H atoms from a dihydride unit and convert a region of H-terminated 3x1 into a H-terminated 2x1 structure. The dynamical process of this conversion and a proposed mechanism will be presented. The unique role of using STM to explore various H-desorption channels and transition states will be discussed.
10:20 AM		EM+SS-TuM-7 Cryogenic UHV-STM Study of Electron Stimulated Desorption of Hydrogen and Deuterium from Silicon(100) E.T. Foley, A.W. Kam (University of Illinois, Urbana); Ph. Avouris (IBM T.J. Watson Research Center); J.W. Lyding (University of Illinois, Urbana) A cryogenic variable temperature UHV-STM has been developed. This design utilizes a novel vibration isolation scheme which provides excellent thermal coupling to a cooling source. The cooling scheme departs from other cryogenic UHV-STMs where vibration isolation and cooling compete with each other. Variable temperature operation from 11 K to 300 K has been demonstrated. Future improvements will enable operation down to 1.5 K. This system has been used to perform low temperature desorption studies of hydrogen and deuterium from Si(100) surfaces. Comparing these results to previous room temperature studies indicates there is no temperature dependence to the desorption in the higher voltage field emission regime. However, in the lower voltage inelastic tunneling regime a strong temperature dependence is observed, with the desorption yield for hydrogen at 11 K being 300 times greater than at 300 K. In the context of Avouris' vibrational heating model¹, these data are consistent with an increase in the Si-H vibrational lifetime from 10 ns at 300 K to 19 ns at 11 K. By similar analysis, the Si-D vibrational lifetime increases from 0.25 ns at 300 K to 0.85 ns at 11 K. ¹T.-C. Shen, C. Wang, G.C. Abeln, J.R. Tucker, J.W. Lyding, Ph. Avouris, and R.E. Walkup, Science 268, 1590 (1995).
10:40 AM		EM+SS-TuM-8 Infrared Spectroscopy of Hydrogen in Silicon: Defects, Platelets and Exfoliation M.K. Weldon, Y.J. Chabal, V.E. Marsico, Y. Caudano, M. Collot (Bell Laboratories, Lucent Technologies); W.B. Jackson (Xerox Palo Alto Research Center); D.C. Jacobson (Bell Laboratories, Lucent Technologies) The profound influence of hydrogen on silicon is exemplified by the remarkable observation that H implanted at doses of 10¹⁶ -10¹⁷ H/cm² can be used to drive the exfoliation of the overlying Si layer. We have previously used infrared spectroscopy in combination with Forward Recoil Scattering and TEM to show that the implanted hydrogen evolves from point defects to form agglomerated vacancy complexes, before stabilizing internal (100) and (111) planes. Simultaneously, H₂ is produced and accumulates inside the extended internal cracks (formed by intersection of these low index planes) and provides the necessary pressure to separate the Si substrate. Since the lattice disruption created by the initial H implantation has been directly correlated to the growth of extended defects, we have investigated the evolution of bound hydrogen as a function of dose, implantation depth and annealing temperature after different H treatments. Importantly, we find that co-implantation of He or Li (with H) causes the formation of initial defects which nucleate these insoluble elements. In turn, this leads to the conversion of H into molecular H₂ at room temperature and the reconversion back to Si-H upon annealing, which is not observed for H alone. We have also studied the evolution of defects produced by i) H and Si co-implantation and ii) H loading by remote plasma source (thermal energy atoms with minimum lattice disruption) which leads to the formation of internal 'platelets'. We will discuss the implications of this work for the understanding of defect behavior in crystalline Si and the physics and chemistry of the exfoliation process itself.
11:00 AM		EM+SS-TuM-9 The Effect of Phosphorus on Hydrogen Desorption from Silicon J.E. Crowell, M.L. Jacobson, M.C. Chiu (University of California, San Diego) Our studies are focused on the effect that dopants have on the surface chemistry of silicon. We have examined the effect that these adatoms have on both the precursor chemisorption behavior and on the hydrogen desorption behavior. This presentation will focus on the effect of pre-adsorbed phosphorus on the adsorption, diffusion, and desorption behavior of hydrogen from silicon single crystal surfaces. We have studied the effect of phosphorus on the etching of silicon by hydrogen, on the desorption kinetics of hydrogen, and on the relative stability of surface hydrides. On Si(100), phosphorus forms both P-P and Si-P dimers. Both of these bonds are shorter than Si-Si bonds, thus introducing strain at the surface, resulting in structural changes at the surface that are dependent on the phosphorus coverage. Furthermore, phosphorus contains a lone pair of electrons and hence does not adsorb hydrogen. We find that pre-adsorbed phosphorus increases the activation energy for desorption, whereas other adatoms such as germanium decrease it. The extent of change in the desorption energy is linear with dopant concentration. On Si(100), phosphorus also stabilizes the dihydride population relative to the monohydride population, perhaps due to structural changes in the Si(100) surface caused by phosphorus adsorption. The hydrogen desorption order for the monohydride also increases above first order in the presence of phosphorus. This change in desorption order is the result of diffusion effects and a reduction in pre-pairing of the hydrogen on Si-Si dimer sites due to formation of P-Si dimer sites. The role of electronic structural changes on the observed hydrogen desorption behavior will also be discussed.
11:20 AM		EM+SS-TuM-10 Effects of Surface Phosphorus on the Kinetics of Hydrogen Desorption from Silane-adsorbed Si(100) Surface at Room Temperatures M. Suemitsu, Y. Tsukidate, H. Nakazawa (Tohoku University, Japan) Hydrogen desorption kinetics from SiH₄/P/Si(100) surface have been investigated by conducting temperature-programmed-desorption (TPD) measurements. When SiH₄ was adsorbed on P/Si(100) with θ_p=0.25 ML, the H₂-TPD spectrum showed a peak shift toward higher temperatures by about 10 degree C as compared to the one from SiH₄/Si(100) surface. Furthermore, the amount of the peak shift increased with decreasing SiH₄ exposure. This implies a higher reaction order for the hydrogen desorption kinetics from SiH₄/P/Si(100) surface. Another possible origin for the TPD peak shift, however, is the increase of the activation energy for the hydrogen desorption. In order to clarify the role of surface phosphorus in these two aspects of desorption, a series of TPD spectra were analyzed in detail. Namely, the desorption activation energy E was obtained by plotting the TPD intensity against 1/T for a given θ value, while the reaction order n was obtained by plotting the TPD intensity against θ for a given T value. As a result, it was clarified that addition of 0.25-ML phosphorus on Si(100) increases both the activation energy E and the reaction order n, from 2.0±0.2 to 2.5±0.2 eV and from 1.0±0.2 to 2.0±0.2, respectively. Thus, the surface phosphorus was found to prevent the hydrogen desorption from Si surface via the increase of (1)the activation energy and (2)the reaction order of the desorption kinetics. This suppression of the hydrogen desorption by surface phosphorus might be responsible for the reduced Si growth rates under PH₃ doping.
11:40 AM		EM+SS-TuM-11 Hydride Reactions on Boron-doped Si(100) G. Hess, B. Gong, P. Parkinson (University of Texas, Austin); S. Jo (Kyung Won University, South Korea); J.G. Ekerdt (University of Texas, Austin) Boron doping influences the growth rate of epitaxial Si(100) films during chemical vapor deposition from Si₂H₆- and B₂H₆ in two different ways. In the low temperature regime, when hydrogen desorption is rate limiting, boron increases the growth rate. Whereas at higher temperatures, when precursor adsorption is rate limiting, a decreasing growth rate has been observed. We are studying these rate limiting steps to elucidate the underlying mechanism causing these effects. Second harmonic (SH) spectroscopy, which is sensitive to the hydrogen coverage, is used to study the adsorption and coadsorption of Si₂H₆ and B₂H₆ on Si(100) under film growth conditions. These measurements yield a reactive sticking coefficient for Si₂H₆ on intrinsic Si(100) of 0.06±0.01 for the temperature range between 740 K and 900 K. Temperature programmed desorption (TPD) measurements are used to probe desorption kinetics. Intrinsic and boron-doped (10¹⁵ - 8x10¹⁹ B/cm³) Si(100) samples with different dopant concentrations were tested. Each sample was exposed to the same 150 L atomic hydrogen dose, which is sufficient to reach saturation (1.33 ML) on intrinsic silicon. A broad TPD feature consisting of two peaks is observed between 500 K and 640 K for boron-doped samples in addition to the Si(100) dihydride (680 K) and monohydride (795 K) peaks. The total peak area of this low temperature desorption feature increases with increasing boron concentration (0.4 ML/5x10¹⁸ B/cm³ to 1.4 ML/8x10¹⁹ B/cm³). Subsurface hydrogen, trapped at boron sites and forming a boron-hydrogen complex, can explain this observation. The formation of such complexes is known to passivate the electrical activity of boron acceptors, which in turn decreases the dopant induced space charge electric field. We observed this passivation in the SH spectra from these boron-doped samples that showed, after 150 L atomic hydrogen exposure, a strong quenching of the boron-related, electric-field-induced SH contribution.