The concept of an "interface dipole" is often abused. When properly defined, a qualitative and quantitative discussion of an interface dipole can yield important information about the (heterojunction) interface. Unfortunately, very often no proper definition is given, and many people do not even realize that without such a definition the concept is meaningless. . . . . . . . . The bottom line is that the concept of interface dipole is best avoided, unless one specifies explicitly which definition is being used, and uses the concept only to identify trends between situations which are sufficiently similar (e.g., same interface orientation) to allow a meaningful comparison.

The above warning issued by A. Franciosi and C. G. Van de Walle from their review article in Surface Science Reports 25, 5 (1996) should be taken to heart by all interested in the electronic properties of interfaces.

F^o_B,n = f_M - c_S ,

(1)

where f_M is the work function of the metal and c_S is the electron affinity of the semiconductor, can be viewed as a zeroth-order theory on the formation of the SBH. Even though the barrier now bears his name, the theory that Schottky advanced has not demonstrated much relevance for real Schottky barriers. Schottky's main contribution, instead, lies in correctly predicting the electric potential and field distribution in the space charge region. A common perception of the Schottky-Mott relationship is that this rule predicts the barrier height in the absence of interface states. Perhaps the best way to see the rationale behind this perception is to derive the Schottky-Mott condition in a "Gedanken" experiment, as illustrated in this figure.

In this exercise, large flat surfaces of metal and semiconductor are assumed to be placed parallel to each other and with a small gap d_gap separating them. An electrical connection is made externally, thus equating the Fermi level of the two crystals ( V_a = 0 ). For a uniformly doped, non-degenerate semiconductor, the total charge per unit area arising from charges in the depletion region is Q_SC = (2ee_S N_D V_bb )^1/2. If the semiconductor has no surface states, as assumed, the space charge, Q_SC, is the only source of charge on the semiconductor surface. The charge on the semiconductor is balanced by a charge of equal magnitude, but of opposite sign, on the surface of the metal. These charges lead to a constant electric field in the gap between the semiconductor and the metal, much like that in a parallel plate capacitor, and a total potential drop of

D_gap = - d_gap Q_SC / e_o .

(2)

Since the electric potential is continuous everywhere, the electric field in the gap is also required to be, according to Fig. 2.1,

D_gap = c_S + eV_bb + eV_N - f_M .

(3)

Combining Eqs. (2) and (3) allows the total band bending to be determined as a function of the gap width, as plotted in (b) of the above figure. As expected, the band bending is zero when the separation is large. And when the gap collapses ( d_gap -> 0 ), one gets what amounts to the Schottky-Mott condition. So, the Schottky-Mott relationship can be viewed as the asymptotic result of the band lineup when a semiconductor without surface states approaches a metal. There is thus some perceived connection between the Schottky-Mott relationship and the absence of surface states. This perception, even though expressed profusely in the literature, is incorrect.

    Some may be quick to point out that the Schottky-Mott relationship is also trivially obtained if the vacuum level outside the metal is lined up with the vacuum level outside the semiconductor. What this entails is for the electric field to vanish in the gap between the metal and semiconductor, which can be accomplished by applying an external bias of ( f_M - c_S - eV_bb - eV_N ) to the semiconductor, as illustrated in this figure.

   With such a bias applied, the semiconductor band bending will be independent of the width of gap and the semiconductor can approach the metal with its CBM already positioned at f_M - c_S above the E_F of the metal. Again, the Schottky-Mott relationship can be inferred as the asymptotic result when the gap becomes zero. Note that this result is always obtained, independent of the density of surface states on the semiconductor. Therefore, the Schottky-Mott relationship has nothing to do with the density of surface states. Rather, it is simply related to the absence of a gap between the metal and the semiconductor. This somewhat disturbing result is of course just an artifact of the analysis model that has been used. The analysis model assumes that the interface dipole is made up of transferred charge between the metal and some pre-existing states on the semiconductor surface. So when the gap vanishes, there can be no interface dipole! In reality, when the semiconductor does come into contact with the metal, chemical bonding will form, new states will be created, charge will flow, and it really doesn't matter how the two materials were brought together in the first place, nor whether or not the semiconductor had surface states before.

   With the concept of the interface specific region (ISR), it is plain to see what the Schottky-Mott relationship really implies about the charge distribution at such a MS interface. It can be shown that the Schottky-Mott relationship is equivalent to

D_ISR = D^hkl_M - D^hkl_S ,

(4)

namely, the interface dipole is the difference between the surface dipole of the metal and that of the semiconductor. This condition is trivially satisfied if the charge distributions on the two original, isolated surfaces are maintained, or frozen-in, at the eventual MS interface. In other words, the Schottky-Mott relationship results from an absence of charge rearrangement or interaction. Since significant charge redistribution at a MS interface usually occurs when new bonds are formed and/or when old bonds are broken, the Schottky-Mott relationship is not expected to give accurate predictions of the SBH at common MS interfaces where interface bonding is unavoidable.