NEA GaAs

For p-type material, the Fermi level within the bulk is very near the top of the valence band, or Ej E — Eq. The electron affinity under these boundary conditions is computed from the identity 0 = X- - (E — Ep), where (E — f) is a positive energy difference Eq this gives X (j) — E = 1.4 —1.4 = 0eV. Zero or negative electron affinity is therefore achieved for p-type material. For the n-type material, Ep lies near E or (E — Ep) 0. Here X = (f) — (E — Ep) (/ — 0 = (/), and 

NEA is not reached. In both cases the real electron affinity at the surface is the same +0.5eV see in both halves of the figure. 

Near-zero electron affinity for p-type GaAs results from a fortuitous match between the work function of Cs and the bandgap of GaAs (both at 1.4 eV) for a case where the Fermi level in the bulk is located below the bottom of the conduction band. This was observed prior to fabrication of the first NEA photocathode, p-type GaAs/Cs [5.66] after construction, the experimentally measured quantum efficiency per absorbed photon was in the order of 0.2 for energies above 2eV. A sharp response cuton at 1.4 eV gave a white light sensitivity of about 500 pA/lm - immediately superior to the classic S-20 surface [5.66]. 

Further lowering of the surface barrier and some improvement in sensitivity of the activated GaAs surface occur when oxygen and Cs are alternately leaked onto the Cs-covered surface to form (CsO) [5.75]. This X-reduction results qualitatively from the lower electron affinity of the substance CS2O, where X-(CsO) ultimately drops to about 0.6 eV for sufficiently thick layers. (See Fig. 5.3, 

This structure of the fully NEA GaAs/(Cs CsO) photoemitter is shown in Fig. 5.10 [5.46]. Note that the GaAs electron affinity is 0.3 eV lower than that found with Cs alone, or — 0.3eV. Further reduction to — 0.8eV would be obtained with a thicker (CsO) layer, but this is neither necessary for NEA nor useful once all energy barriers are overcome. 

---

Negative electron affinity (NEA) GaAs reflection photocathodes which exhibit sensitivities as high as 2150 pA/lm. 

NEA GaAs/InGaP/GaP transmission photocathodes which exhibit quantum efficiencies as high as 15% at 8500 X. 

Considered first in this chapter are the physical principles of operation applicable to both classes of photoemissive surface, to the extent to which they are understood. The operation and construction of classical devices are then examined in detail, using two extreme cases, (CsSb) and (AgCsO) -S —1, as examples. NEA devices are then considered. These are discussed in somewhat greater depth than classical devices inasmuch as these recently developed detectors are less likely to be familiar to the user. One example of NEA operation, NEA GaAs, is the simplest structure and is discussed in detail for both RM and TM modes. Other IR-sensitive emitters, including those using layers of complicated quaternary compound semiconductor alloys such as InGaAsP, are then briefly summarized. The chapter is concluded with a summary of device-to-device trade-offs in classical and NEA devices. 

Even for electrons with sufficient energy at the surface, the probability of actual emission is always less than unity. Escape varies with the type of material, its crystal orientation, and even with the specific conduction band to which the carrier has been excited. For NEA GaAs, for example, the escape probability from the 111-B (As) face is about 2.5 times larger than that from the 111-A (Ga) face [5.46], and escape from X-states is about twice as probable as escape from F-states [5.47]. Little of this kind of detail is known about escape from classical materials. Surface escape factors for both NEA and classical materials are nearly impossible to model, much less predict, and the search for materials with both... 

Fig. 5.9a and b. Development of near-NEA GaAs/Cs from the band structure of its constituents. In (a), Ga As and Cs are shown separately in (b). both p-type and n-typc GaAs are shown after coating with a thin layer of Cs. In both sectionsof part (b), the value of p at the surface is pinned to and the surface values of Ec and y sf identical for n-typc and p-typc material. The electron affinity for p-type Ga As/Cs is near zero, while that for an n-typc sample is near <- = 1.4cV. The position of p in the hulk, determined by dopants of energy not shown, is the only difference between the construction of the two sections. (Although the Fermi level is shown here pinned to ss in other models y or Ef-may be pinned to a surface state energy instead) (adapted from [5.13])... 

Fig. 5.10. Fully NEA GaAs/(CsO) according to the heterojunction model. For the optimum (CsO) thickness, the work function of (CsO) is 1.08 eV, which gives a bulk GaAs electron affinity of-0.34 eV. The conduction band discontinuity, common in most coated IIl-V surfaces, is hidden for the 111 B face of GaAs. Note the slight (0.1 eV) band bending, indicating that Ess differs from its value on pure Cs-coated material. The Fermi level at the surface varies also with the choice of activated crystal face [5.46]...

Before considering other possible TM or RMIR N E A photoemitters, we discuss in this subsection the fabrication of NEA GaAs and the general factors required to optimize RM mode photocathodes [5.46,47, 83, 89 -91], plus the additional special constraints of TM mode fabrication [5.14, 15, 89, 91, 92-96]. 

Production of NEA GaAs requires cleaning and cesium-activation procedures [5,97]. All steps are performed under ultrahigh vacuum (oxygen-free) conditions after a system bakeout. 

---