I. CONTROL OF SCHOTTKY BARRIER-HEIGHT
I.1. Introduction and Background
As the demand for more and better functionalities in electronic and photonic devices continues to rise, the search for better materials and better structures also intensifies and widens in its scope. A general trend for the development of future electronic devices has been to integrate an astronomical number of minuscule units together, each consisting of a large variety of materials. When materials with widely different atomic structures and electronic structures are tightly integrated into a small dimension, the control of the electrical integrity of the conglomerate becomes a formidable issue and is of paramount importance to the overall performance of the device. Carrier transport across macroscopic solid interfaces is governed by a parameter generally referred to as the “band-offset” of that interface. Depending on the specific material interface systems, this parameter is also known as the “Schottky barrier height” (SBH), the “molecular energy level alignment”,[3, 4] etc. The ability to control the magnitude of this barrier height is crucial for the advancement of future electronic devices to higher functionality and smaller physical dimensions. Presently, our understanding of the formation mechanism of the barrier height falls well short of ideal. In the literature, the formation of interface dipole has predominantly been discussed in terms of interface gap states and the charge neutrality level (CNL). However, the recent discovery that the magnitude of the interface dipole at metal-semiconductor interfaces can be accounted for by interface bond polarization alone,[6-9] along with the well-known problems of the CNL concept,[10, 11] has rekindled discussions on the basic band-offset mechanism. It is certainly important at this stage to investigate the band-offset mechanism to the best of our abilities. From the perspective of meeting technological challenges, it is perhaps even more important to look beyond the present debate and actively search for empirical or semi-empirical methods to control the barrier height.
I.2. Existing Strategies for Barrier Height Control
Existing schemes to control the barrier height can be roughly divided, according to the approaches used, into three groups. The simplest approach is to make use of the empirical dependence of the magnitude of the SBH on the interface structure[2, 12, 13] and the processing condition.[14, 15] These dependences not only test the validity of the SBH models but the underlying reason for these dependences may also point to new ways to tune the barrier height. Amidst a myriad of sundry correlations experimentally observed, the most intriguing and poignant parameter which significantly influences the SBH is the deposition temperature.[14, 16-21] Specifically, deposition of metal at cryogenic temperatures on compound semiconductors frequently led to an increase in the n-type SBH, when compared with junctions formed at room temperature. Most spectacularly, deposition of metal at 77K was shown to lead to an apparent unpinning of the Fermi level position at ZnSxSe1-x interfaces. A second approach involves the insertion of an inorganic interface layer between the semiconductor and the metal, in order to modify the SBH. This has been accomplished by the introduction of a very thin layer, typically sub-monolayer to a few monolayers, of chalcogen (S, Se, ..),[22-29] Sb,[30, 31] nitride, and oxide,[33-35] to the interface, or by slightly oxidizing the compound semiconductor surface before the metal deposition.[36, 37] Changes in the SBH of GaAs and InP by as much as 0.4eV have been observed. However, at some interfaces, it was discovered that the large SBH changes were observable only with some measuring techniques, but not with others.[33, 35] The insertion of a thin layer of a semiconducting material between metals and a semiconductor substrate led to changes in the SBH, which have been more consistently measured by different techniques.[32, 38-43] The third approach to SBH modification makes use of the dipole moment of organic molecules. The inclusion of a submonolayer or monolayers of polar organic molecules at the metal-semiconductor interface, either by grafting self-assembled monolayer (SAM)[44-52] or by applying Langmuir-Blodgett films,[53, 54] provides conceptually the most straightforward method to vary the interface dipole and, along with it, the barrier height. The change in the barrier height, , is expected to be
where is the dipole moment of the molecule, is the number of molecules per area, is the effective dielectric constant of the molecular layer, and is the average angle the molecules make with the surface normal. Changes in the barrier height consistent with the prediction of Eq. (1) have been observed, although depolarization effect and other effects have also been observed.
I.3.a. Challenges for Metal Contacts to Inorganic Semiconductors
Since metal connections are required for virtually all solid-state electronic and optoelectronic devices, technologies for low-resistance ohmic contact need to be developed for a wide variety of materials. The two common strategies for semiconductor contacts are heavily doping the interface semiconductor region and using metals with as small a SBH as possible. The Fermi level pinning phenomenon has limited the range of available SBH’s for most semiconductors and reduced the effectiveness of the second option. Therefore, it is not surprising that those semiconductors with difficulties in dopant incorporation and/or dopant activation are often the ones with ohmic contact issues. The most well-known of the problematic semiconductors are the p-type GaN,[56-58] the p-type ZnO,[59-62] and the n-type Ge,[27, 63-65] for which a successful contact strategy very likely will center around the ability to modify the barrier height, using methods discussed above. For these semiconductors, and generally for all solid-state materials, the identification of the important parameters that control the barrier height and the development of reliable technologies to take advantage these parameters are needed for improved device performance.
Another area where the ability to tune the SBH is in great demand is in the migration to high permittivity (high-K) gate dielectrics from silicon oxide of the silicon ULSI industry, which appears inevitable beyond the 45-nm generation devices. Beside a high dielectric constant, the gate material needs to have reasonably large offsets with both bands of the Si, in order for the device to work successfully. This very consideration on band-offsets[9, 66-68] is an important factor contributing to the recent, nearly unanimous identification of hafnium oxide/nitride as the primary candidate for high-K dielectric applications. In addition, a metallic gate, rather than polycrystalline Si, should be used with high-K dielectric, to avoid (lower-k) oxide reaction. One philosophy for metal gate is to choose a metal with a work function which matches roughly the mid-gap point of the semiconductor. However, to be able to maintain the threshold gate voltage for the field effect transistor at a convenient voltage, especially at scaled-back power supply voltages, it is desirable to have separate Fermi level positions for gates on n-type and p-type channels.[71-74] For this purpose, one needs to control the Schottky barrier height between the metal gate and the high-K dielectric. A satisfactory solution to the metal-gate issue is still actively being sought.
I.3.b. Challenges for Metal Contacts: Organic Devices and Molecules
Organic/polymeric materials have already found many applications in electronics, photonics, and magnetics, and they are being explored for many more potential applications. With the expected introduction of organic material into highly integrated devices, the role played by interfaces on the electronic transport becomes more vital and can dominate the overall performance of the entire device. Of all organic interfaces, the metal-organic interface is probably the most common and the most important,[75, 76] in terms of applications. However, an examination of the literature reveals that not only is the conduction mechanism at metal-organic interfaces not well understood, but also the fabrication of these interfaces is problematic for very thin organic layers. It is well known that the electrical and chemical properties of a metal-organic interface depend on the “order” of its fabrication. Interfaces formed by the thermal evaporation of organic molecules, or the spin-coating of a polymer layer, on a metal surface (organic-on-metal) tend to have very consistent properties, often with negligible interface electric dipoles. The deposition of metals, by thermal evaporation or sputtering, onto an organic material (metal-on-organic), on the other hand, creates metal-organic interfaces with inconsistent, inhomogeneous, and leaky electrical conducting properties.[78, 79] The problem with metal contact is particularly acute when metallic layers are deposited on very thin organic layers such as SAM layers, Langmuir-Blodgett films, etc. The problem with metal-on-organic interfaces has largely been attributed to the in-diffusion and reaction of incoming metal atoms, particularly with large kinetic energy from the hot evaporation source or the high-energy sputtering process, with the organic molecules.[80, 81] Because of these problems, studies on the electronic transport across a single layer of molecules, or through a single molecular wire,[82-85] have relied on specially designed “soft” contact technologies. Well-known soft contact technologies include such ingenious inventions as “controlled break junctions”,[86-88] electro-migrated junctions, deposition through nanosieves,[83, 90] nano-particle deposition,[91-96] electrochemical methods,[97-102] lift-off float-on (LOFO) method,[51, 103] conducting-probe atomic force microscope (CP-AFM),[104-108] scanning tunneling microscopy (STM),[109-114] crossed-wire contacts,[115, 116] mercury droplet contacts,[117-120] micro-contact printing,[121-126] and nanoscale transfer printing (nTP).[127, 128]
With no metal deposition step, most of the specialized soft-contact methods described above do not encroach on the organic molecules and have enabled successful studies of the electronic transport properties across single molecular junctions. However, none of these laboratory inventions has broad enough applicability to represent a technological solution to the contact problem on organic materials. One seeming exception is the printing (nTP) technique, which is very suitable for manufacturing but requires specific chemistry and therefore is limited in both the type of contacts that can be formed and the kind of SAM molecules that can be studied. Therefore, a general soft contact technology, which allows any metal to be deposited on any type of molecular layers, is still in high demand. The desire for a one-for-all type of technique suggests a physical (rather than chemical) deposition process. An indirect thermalization deposition (ITD) technique,[52, 85, 130] with the evaporation taking place in a vacuum chamber back-filled with inert gas (10-3 torr) at a cryogenic temperature and with the substrate facing away from the source, has produced successful organic contacts recently. Through collisions with inert gas molecules, a small fraction of the evaporated metal atoms is thermalized to the ambient temperature and diffuses in inert gas to reach the substrate. Robust contacts have been demonstrated on SAM layers with this technique. Because of the small deposition rate and the time delay for metal atoms to reach the substrate, the ITD technique is not suitable for manufacturing, nor does it offer atomic level control and co-deposition capability.
II. BASIC IDEAS BEHIND THIS PROJECT
II.1. Motivation from Literature and Own Previous Work
Of all methods designed or discovered to change the magnitude of the SBH discussed above, three are particularly poignant and efficacious: (a) deposition temperature, (b) chalcogen interface layer, and (c) organic interface dipole layer. A careful examination of the various implications of the experimental data in these previous investigations suggests several exciting new directions that should be explored for the benefit of the science and technology of the SBH. These are briefly described in this section.
The role played by the deposition temperature in affecting the SBH has not had a satisfactory explanation. Structural studies suggest that the metal layers deposited at cryogenic temperatures were more uniform than that deposited at room temperature, although little difference was seen in the interface roughness.[20, 21] It should be pointed out that only the temperature of the substrate was varied in the previous studies,[14, 16-21] with no attempt to control or curb the kinetic energy of the hot metal flux. To shed more light on the kinetic processes and the atomic structure of interfaces formed from metal vapor flux, it seems that controlling both the semiconductor temperature and the metal flux temperature is necessary. The ITD technique seems ideally suited for such a study. In addition, most of what can be accomplished by the ITD method seems achievable by colliding the hot metal flux only once with a cool inert molecular beam, in an “instant cool beam” (ICB) technique recently demonstrated in our lab. This novel technique, with improved control over the deposition than the ITD, seems very attractive for the development into a versatile and higher-throughput soft-contact technology.
In contrast to the air of mystery surrounding the role played by the deposition temperature, the mechanisms by which the other two methods (involving inorganic and organic interface layers) modify the SBH are thought to be understood. The introduction of Group VI elements (S, Se, Te,..) at the clean Si(100) surface has been proposed to remove surface dangling bonds, via the valence mending mechanism, and allow the ideal Schottky-Mott relation to take effect:
where is the n-type SBH, is the metal work function, and is the semiconductor electron affinity. The Schottky-Mott relation is expected because the interaction between the metal and the passivated surface is thought to be small and therefore the charge transfer, as well as the interface dipole, is expected to be small. It is known that, with the presence a chalcogen surface layer, the semiconductor electron affinity is significantly increased due to the positive dipole formed between the chalcogen and group IV or III-V semiconductors. (A positive dipole in this document is defined as one with a positive charge toward the semiconductor bulk and a negative charge toward the surface or metal.) That being the case, it seems that the electron affinity of the adsorbate-covered semiconductor, rather than that of the clean semiconductor surface, should be used in Eq. (2), leading to a result in agreement with the observed decrease in the n-type SBH.[28, 29] One should mention that the actual experimental SBH changes have not been interpreted in terms of the Si-S or Si-Se dipole previously, but only as a result of removing surface states.[28, 29, 134] Three important deductions can be drawn naturally from this picture: (i) The magnitude of the change in the semiconductor work function due to chalcogen-Si chemistry is a very large effect, ~0.4eV. (ii) To observe the full effect of the “interlayer”, the metal should be deposited on the passivated semiconductor surface as “gently” as possible to avoid chemical interaction and interfacial charge transfer. This may be conveniently accomplished by using ITD (or the new ICB technique described below). (iii) If a chemically stable surface can be formed with adsorbates that donate electrons to the semiconductor surface, such as Mg, Ga, etc., a negative surface dipole and an increase in n-type SBH should result. A detailed study of the dependence on the chemical specificity (or electronegativity) of the interlayer element may reveal the fundamental SBH mechanism directly. The use of surface adsorbates other than Group VI chalcogens as an interlayer in SBH investigations is largely an unexplored territory. With data on impurity-stabilized silicon surface structures widely available for adsorbates ranging from alkali-metals all the way to halogens, a systematic study into the SBH formed with group I through group VII elements as interlayer is already within reach. Such a study is expected to uncover a great deal about interface charge transfer and, furthermore, may develop into, or identify, techniques to directly control the SBH on a significant scale.
The electrostatic effect of a layer of dipolar organic molecules is straightforward to predict, as Eq. (1) suggests. On a free surface, with a net layer dipole moment perpendicular to the substrate, organic molecules can produce a substantial shift in the work function of metals[44, 135, 136] and in the electron affinity and work function of semiconductors.[137-142] To first order, these changes are due to rigid shifts in the electrostatic potential across the adsorbed molecular layers. Because molecules can be synthesized with the same binding group and the same backbone, but with different functional groups, a variation in the group electronegativity of the functional group allows the molecular dipole to be systematically changed, even to the extent that the sign of the dipole can be reversed. Systematic studies of SAM molecules based on single carboxylic acid binding group (benzoic acid derivatives[135, 143, 144]) and dual binding groups (dicarboxylic acid derivatives[52, 145-147]) on GaAs have shown that molecular dipoles can control metal/semiconductor junction even when the molecules are neither well organized, nor close-packed, but are in incomplete layers with a significant fraction of pinholes.[51, 146, 148-151] Since the tunneling efficiency across a molecular layer is small compared with thermionic emission over intimate metal-GaAs interface, essentially the entire current transport in a metal-(porous molecular layer)-semiconductor (MpmS) junction can be attributed to pinhole conduction.[52, 152] The ability of a partial molecular layer to increase the SBH in the pinholes is easily understood to originate from the pinch-off effect, a phenomenon well studied for inhomogeneous SBH.[153-155] This molecular effect depends to a large extent on the difference in barrier height for electron transport between adjacent domains, the molecular coverage on the surface, the concentration, the sizes of the pinholes, and the doping level of the semiconductor. The reason that a partial molecular layer can lower the SBH through pinholes is more intriguing, and has been tentatively attributed to enhanced edge conduction through pinholes, aided by the dipole field of the molecular layer.[52, 152] To clarify this mechanism and also to take advantage of the demonstrated ability of partial molecular dipolar layers to lower the SBH and improve conduction at ohmic contacts, it seems beneficial to be able to deliberately create MpmS junctions that are porous. Two variations on this MpmS theme can be identified and are within the experimental capabilities of the PI and his collaborators. It will be explained below how to first self-assemble “porous metal” template-islands and then fill in the gaps with dipolar molecules AND on how to first artificially create “porous molecular layers’ and then fill the gaps with metals. A lateral scale of 5-100 nm for the inhomogeneous MpmS junction can be achieved in either of these schemes.