A new class of electron emitters

N. A. Soboleva

All-Union Institute of Scientific and Technical Information ( VINITI)

Usp. Fiz. Nauk 111, 331-353 (October, 1973)

This review reflects the status (in early 1973) of the problem of producing photoemitters andsecondary and field emitters based on a new principle, that of obtaining semiconductor structureswith negative electron affinity (ΝΕΑ). The energy level scheme of the surface region of a semi-conductor with ΝΕΑ is examined, as well as the conditions for its realization. The main theoreticalconcepts of the mechanism of the emission from ΝΕΑ emitters are developed, and the differencebetween their characteriistics and the corresponding characteristics of ordinary emitters are indi-cated. The technology is discussed of the production of ΝΕΑ photocathodes for the visbile andinfrared region of the spectrum, based on III-V semiconductor compounds (principally GaAs) andtheir solid solutions. Aspects of the production of semitransparent ΝΕΑ photocathodes and sec-ondary emitters operating in transmission are noted. The principal parameters of the experimentaland commercial ΝΕΑ photocathodes and secondary electron emitters are given. Data are reportedon the use of surfaces with ΝΕΑ in field-emission film cathodes, the operating principle of injectionand optoelectronic field emission cathodes is described, and their present-day parameters are given.

Practically all of the technical photoemission andsecondary-emission materials that have been used up tonow in photoelectric devices (vacuum photocells, photo-multipliers, electron-optic converters, transmittingtelevision tubes) were discovered in the period from1930 to 1955 through empirical searches, technologicaltests, and chance findings. The first attempts to studysystematically the properties of emission materials andto develop theoretical views of how they work were un-dertaken in 1939-1940. These studies were renewed atthe end of the forties and intensified, having started togrow anew in connection with the creation of a generaltheory of semiconductors. These studies, mainly onthe photoemission properties of the alkali antimonides,resulted in formulation of the main requirements thatefficient emitters of photo- and secondary electronsmust satisfy/1"33

The process of any non-equilibrium electron emissionconsists of three stages: 1) excitation, 2) transport tothe surface, and 3) escape of the electrons into thevacuum. In photoemission, the first stage is determinedby the optical properties of the material, and in secondaryemission, by the laws of interaction of fast primaryelectrons with the solid. The latter stages differ littlein photo- and secondary emission.

The transport of excited electrons is usually charac-terized by the effective escape depth, i.e., the mean dis-tance that they can travel while yet remaining able toescape. The greater the escape depth of the electronsis in comparison with their excitation depth, the moreefficient the emitter is. In the usual semiconductor ma-terials, hot electrons participate in emission, and theyremain able to be emitted until they dissipate part oftheir excess energy and are left in an energy level belowthe vacuum level. The scattering length of hot electrons(which is equal to their escape depth) is determined bythe nature of their interaction with the solid. If the

electrons can dissipate energy by exciting valence elec-trons of the semiconductor (i.e., in impact ionization,which is accompanied by generation of electron-holepairs), then the scattering length does not exceed themean free path, which is 10-20 A.

If such interactions are energetically impossible —for photoelectrons excited near the threshold of thephotoeffect, they are impossible if the materialpossesses an electron affinity11 smaller than the widthof the forbidden band, then the main form of dissipationof energy of hot electrons will be interaction with opticaphonons and lattice defects. Then the escape depth ofhot electrons depends on their energy, and it can exceedby factors of tens the mean free path, being as much as-150-300 A. In fact, even in this most favorable case,the escape depth of hot electrons proves to be substan-tially less than the depth of optical absorption, especiallynear the threshold of the photoeffect, where the absorp-tion coefficient is small (a ~ 104 cm"1; l /α ~ 1

Semiconductors having a small enough electron af-finity (at least having % a < £g) and any further decreasein 8 a will improve the transport properties, and thusmust improve the emission efficiency of the material.We can say that the electron affinity of semiconductorsis the main parameter that determines their emissionefficiency.

Figure 1 shows the energy diagrams of the surfaceregions of semiconductors having p- and η-type con-duction. As we see, the bending of the bands at the sur-face has the result that the effective electron affinity(i.e., the energy spacing from the vacuum level to thebottom of the conduction band in the bulk of the materialbeyond the narrow band-bending region) declines forp-type semiconductors, but increases for η-type, ascompared with weakly-doped semiconductors having flatbands. In addition, since the bands are bent upward in

726 Sov. Phys.-Usp., Vol. 16. No. 5, March-April 1974 Copyright© 1974 American Institute of Physics 726

FIG. 1. Energy diagramsof the subsurface region ofsemiconductors having n-and p-type conduction.

η-type semiconductors, this creates an additional bar-rier for conduction electrons at the vacuum boundary.Conversely, the bending of the bands in p-type semi-conductors favors emission. Hence p-type semicon-ductors are better emitters than n-type.

As studies have shown, all of the known technicalphotocathodes that can be classified as "efficient photo-emitters" (having quantum yields >0.1 at relativelyshort distances from the threshold) confirm theseideas: they are semiconductors having low electronaffinities and p-type conduction.

The most sensitive of the photocathodes, which be-long to the class of A*BV compounds, is a muItialkaliphotocathode of composition K, Na, Cs-Sb. It is char-acterized by the following parameters: a mean integralsensitivity 200-250 μΑ/lumen, and in the modern, per-fected modification, up to 450 μΑ/lumen and even 600μΑ/lumen with optical amplification using total internalreflection. The peak of the spectral characteristic lies1.0-1.5 eV from the threshold, and the quantum yield atthe peak is as much as 0.2-0.4. The threshold wave-length is as long as 940-960 nm for the best specimens.The quantum yield is 3-4% in the 800 nm region. Thethermal current density at room temperature is ~10~15

A/cm2.

The secondary-emission properties of materials arealso mainly determined by transport of relatively slowsecondary electrons, and the conditions that ensure ahigh secondary-emission coefficient practically coincidewith those for high photoemission efficiency, but withoutthe restrictions on optical properties and on the widthof the forbidden band. Photomultiplier dynodes are mostfrequently oxides of magnesium or beryllium and photo-sensitive CsaSb films. At the peak of the photoemissioncharacteristic (with a primary voltage of 0.8-1 kV), thesecondary-emission characteristic of these materials isas great as 10-15.

The formulation of clear-cut ideas on the mechanismof emission by semiconductors has opened the path topurposefully influencing the properties of the materialsin order to raise their emission ability. As we see fromthe abovesaid, they fundamentally amount to developingways of reducing the electron affinity of semiconductors.

The following evident relationship exists between thephotoemission threshold h o and the work function φ ofa semiconductor:

Av0 = = etp • (1)

Here &$ is the energy gap between the top of the valenceband and the Fermi level.

We get from Eq. (1)

der to reduce the electron affinity S a of a materialhaving a fixed width of forbidden band: a) reduce thework function φ , and b) reduce 6g . The ways of reduc-ing the work function have been well studied; in orderto do this, one coats the surface with a film of electro-positive atoms (e.g., cesium) or molecules having alarge dipole moment (BaO, CsF, etc.). One can attainthe other goal of decreasing 6g by doping the semi-conductor with an acceptor impurity: the higher the de-gree of doping, the lower the Fermi level is dropped.

The limiting manifestation of these efforts is tolower the electron affinity to zero or even to a negativevalue. In order to do this, as we see from (2), we mustincrease the acceptor concentration to a level such thatthe Fermi level coincides with the top of the valenceband (eg =0). If here we lower the work function at thesurface of the semiconductor to a value equal to or lessthan the width of its forbidden band, we get

, %a = βφ - %e < 0. (3)

The Dutch physicists Scheer and van LaarC4] firstproduced a photoemitter having a zero electron affinityin 1965.

To reduce the work.function, they used a nearly-monatomic cesium film. When the latter is adsorbedonto a clean surface of a material, the work function ofthe latter acquires at the minimum a value approxi-mately equal to the ionization energy of the adsorbedcesium atoms (~1.4 eV). Gallium arsenide proved tobe the most suitable material for making a photoemit-ter having zero electron affinity: it has gg= 1.4 eV,straight bands that make possible high optical absorp-tion near the edge of the intrinsic band, and it easilydissolves an impurity (e.g., Zn) in very high concentra-tion (>1019 cm"3). The first photocathode was preparedaccording to this system from GaAs with a Cs film on asurface cleaned by cleaving a single GaAs crystalin an ultrahigh vacuum. It immediately showed an in-tegral sensitivity of 500 μΑ/lumen, which exceeded by afactor of two the sensitivity of the most efficient ofordinary photocathodes, and it had a photoemissionthreshold that coincided with the long-wavelength op-tical absorption edge (hi>0= %g).

The absence of a potential barrier at the surface ofa semiconductor having zero or negative electron af-finity fundamentally changes the nature of emissionprocesses. Figure 2 shows the emission diagrams andthe energy distribution of emitted electrons for semi-conductors having positive and negative electron af-finities. As we see, when there is no potential barrier,not only hot electrons can participate in emission, butalso thermalized electrons, i.e., electrons that havedropped to the lower levels of the conduction band. Thisfundamentally changes the mechanism of transport ofthe excited electrons: it is converted from transport of

FIG. 2. Diagrams ofelectron emission fromsemiconductors havingpositive (a) and negative(b) electron affinity.

(2)

We can easily see that we should do the following in or-

727 Sov. Phys.-Usp., Vol. 16, No. 5, March-April 1974 N. A. Soboleva 727

hot electrons into diffusion of minority carriers in thesemiconductor.

The lifetime of thermalized electrons, which is gov-erned by recombination, is 10"8-10"9 sec for AIHBV-type semiconductors, and ~10"6-10"5 sec for Si. Thatis, it exceeds by a large factor the time for thermali-zation of hot electrons (10~14-10~12 sec). The escapedepth of the photoelectrons coincides with the diffusionlength of electrons in the semiconductor, which usuallyconsiderably exceeds the escape depth of hot electrons,and it can exceed the optical absorption depth l / α evennear the edge of the intrinsic band.

Achievement of a negative electron affinity permitsmost of the excited photoelectrons to escape into thevacuum, and this leads to very high quantum yields, evenin the immediate vicinity of the photoeffect thresholdhiO= 8g.

At high doping levels of the semiconductor, which isone of the conditions for achieving zero electron affinity,the width w of the band-bending region at the surfacedoes not exceed 50-100 A. Thus, we can assume that allof the radiation is absorbed in the bulk of the semicon-ductor beyond this region. The electrons that are excitedat a distance of the diffusion length from the surface arequickly thermalized. As they diffuse to the surface, theyenter the band-bending region. An electric field isconcentrated there and accelerates them toward the sur-face. In the band-bending region, the thermalized elec-trons again become hot electrons. That is, the mech-anism of phonon energy loss acts on them, and reducestheir probability of escape. At a high enough concentra-tion of doping impurity, the width of the band-bendingregion does not exceed substantially the mean free pathof the electrons for phonon scattering, and the electronspass through it with insignificant losses. By loweringthe work function to the point of negative electron af-finity, one can reduce the "detrimental" part of theband-bending region, i.e., the part of it where the bot-tom of the conduction band lies below the vacuum level.

The spectral characteristics of photoemitters havingnegative electron affinity (ΝΕΑ photoemitters) differfrom ordinary photocathodes in their uniformly highquantum yield over a broad spectral range up to thephotoemission threshold, which is determined by thewidth of the forbidden band of the material. That is, theyare distinguished by a steep rise in spectral sensitivitynear the threshold. ΝΕΑ emitters of secondary electronsare characterized by considerably higher secondary-emission coefficients for high-energy primary electrons,which penetrate to great depths into the material.

Since most of the excited electrons can thermalizebefore escaping into the vacuum, ΝΕΑ emitters, bothphoto- and secondary emitters, differ from ordinaryemitters in their considerably narrower energy spectrumof emitted electrons and smaller mean value of initialenergy (as we can see from Fig. 2).

James, Moll, and Spicert5"73 have developed a quan-titative theory of action of ΝΕΑ photoemitters based onthe example of gallium arsenide coated with a film ofcesium.

Figure 3 shows a simplified band diagram of GaAs.The optical transitions in this material are direct elec-tronic transitions. That is, they occur with conserva-tion of the wave vector, as shown by the arrows in Fig.3. At a photon energy hv< 1.75 eV, the excited electrons

FIG. 3. The band struc-ture of GaAs. a-Directelectronic transitions forhi><1.75eV;b-forhK>1.75eV.

-2

(m) (000)

are thermalized in the Γ\ minimum of the lower con-duction band. When ην > 1.75 eV, part of the electronsis excited to levels above the Xi minimum, and areeither thermalized into it, or are scattered from theXi minimum into the Tl minimum. The distribution ofoptically excited electrons between the ΓΊ and Ximinima can characterize the coefficients F r and Fxas functions of he, and they are calculated from theband structure of GaAs. When hf>2.3 eV, a consider-able fraction of the electrons is excited near the sur-face (the absorption coefficient a is large here), andthey escape into the vacuum as hot electrons.

Let us assume that we can neglect absorption of ra-diation in the band-bending region and in the surfacefilm of cesium and that direct recombination of elec-trons from the Xi minimum to the valence band doesnot occur. Then we can describe the motion of elec-trons thermalized in the two minima to the surface(along the ζ axis normal to the surface) with the fol-lowing diffusion equations:

τ Χ Γ(4)

(5)

Here ζ is the distance of the emitter from the surface,the D are the diffusion coefficients of Γ and X elec-trons, TV is the recombination lifetime of Γ electrons,and Txr is the relaxation time for scattering of elec-trons from the X to the Γ minimum. The term ηχ/τχΓsimultaneously determines the rate of generation of elec-trons in the Γ minimum by scattering from the X min-imum and the rate of loss of electrons from the X min-imum.

The term containing the exponential factor in bothequations determines the rate of optical generation ofelectrons. J is the number of photons incident per unitsurface of the emitter, and R is the coefficient of re-flection of the material.

Simultaneous solution of these equations permits usto determine the flux density of electrons arriving at theboundary of the band-bending region. Upon introducingthe factors Ρχ and P r , which characterize the prob-ability of escape of electrons into the vacuum, we getthe following expressions for the quantum yield ofphotoemission (i.e., the numbers of emitted electronsper incident photon):

(6)

(7)

Here L = VDT= τ/μτ^Τ/β) is the diffusion length of theelectrons.

We see from expressions (6) and (7) that one can getthe largest quantum yield in an NEA-photocathode when

728 Sov. Phys.-Usp., Vol. 16, No. 5, March-April 1974 N, A. Soboleva 728

the diffusion length of the electrons is so large that itsatisfies the inequality

(8)

while the probability of escape of the electrons at thesurface Ρ —• 1. In the limit, the quantum yield near thethreshold of the photoeffect (Yr) approaches( l - R ) P r . The diffusion length L r of the minority car-riprs is a bulk property of the emitter material. Itssize depends on the perfection of structure of the semi-conductor, which is determined by its method of prep-aration. Everything that reduces the recombinationtime of electrons or lowers their mobility (extremelyhigh doping levels, presence of contaminating impuri-ties, lattice defects, dislocations, and precipitates)also reduces the diffusion length Lp. Thus, an increasein the acceptor concentration from l x 1019 to 4x 1019

cm"3 decreases Lf from 1.6 to 1.0 Mm.:i4>203 The re-quirement of a high concentration of acceptors forlowering the Fermi level in the bulk and for decreasingthe width of the band-bending region at the surface con-tradicts the requirement for a long diffusion length ofelectrons. Hence, an optimal doping level that permitsa reasonable compromise between these parameters ischosen for each material.

In the first photoemitters based on single crystals ofGaAs doped with zinc at concentrations l-3x 1019 cm"3,Lp was as much as 1.2-1.6 μΐη.1 6 ' 1 7 ' 2 0 1 Increases indiffusion length have been gained in different ways: byperfecting the methods of growing single crystals ormethods of preparing epitaxial GaAs films, by reducingthe doping level with compensation for the widened band-bending region by a more substantial reduction in thework function, e.g., by coating the surface with an ac-tivated film instead of cesium. Such a film can be ob-tained by successive (or simultaneous) treatment withcesium and oxygenC8>15>23] or by using films of othercompounds having high dipole moments, e.g., CsF,CsOH, etc. 1 1 1 ' 1 2 ' 1 3 ' 2 1 3 In the best modern photocathodesbased on GaAs doped with zinc at a concentration ~5x 1018 cm"3, a diffusion length of electrons of as muchas ~6 Mm has been obtained.11303

It has been possible to get larger diffusion lengths byusing as the dopant in GaAs silicon, germanium, ormanganese instead of zinc.C 2 6'2 7 ] A value L r ~ 5-7 Mmhas been obtained in epitaxial gallium arsenide dopedwith Ge at concentrations 5xlO1 7-2x 1018 cm"3.

The probability of escape of electrons from the sur-face of a photoemitter is determined by their scatteringin the subsurface band-bending region and in the activat-ing film that lowers the work function of the cathode. Inorder to increase P, one must reduce the width of theband-bending region, or as it were, the width of its"detrimental" portion (by more substantial reductionof the work function) and one must also use optimal ac-tivating coatings of small thickness. Reduction of thework function, i.e., increase in the difference %g—e<p(the absolute value of the negative affinity) must have itsreasonable limits, since the thermoemission of thephotocathode increases with decreasing φ . BellCl8'193

has calculated all the possible components of thethermoemission current for a typical GaAs-Cs-Ophotocathode and other A ^ I B V compounds. The theo-retical relationship that he got between the specificthermocurrent and the work function φ at the surfaceshows that the thermoemission of GaAs-Cs-O isinsignificant at zero electron affinity (10~19-10~18 A/cm2

under the crudest assumptions). However, it increasesto 10"16-10"15 A/cm2 as the work function is reduced to1.2 eV (although even here it remains at the level of thebest ordinary modern photocathodes).

One can reduce the energy losses in the band-bendingregion by narrowing this region, not only by increasingthe acceptor concentration NA, but also by decreasingthe amount of bending Ag of the bands, as determinedby the position of the Fermi level with respect to thetop of the valence band at the surface of the material.The width of the band-bending region, which must notsubstantially exceed the scattering length for hot elec-trons, is determined by the expression

0)

where e is the dielectric constant of the material, andAg is the amount of bending of the bands, as determinedby the distance from the top of the valence band to theFermi level at the surface. The smaller Ag is, thesmaller the doping level NA that permits one to getthe required value of w, on which the escape probabilityΡ of electrons depends.

A number of studiest 3 5"4 0 ] has discussed the problemof the position of the Fermi level at the surface ofGaAs. Measurement of the height of the Schottky bar-rier at the boundary of GaAs and a thick Cs film indi-cates stabilization of the Fermi level at a distance of~0.6 eVt372 from the top of the valence band.

The density and distribution of surface levels thatstabilize the Fermi level at the surface of gallium ar-senide coated with an adsorbed film are determined bythe nature, and perhaps the thickness, of the adsorbedfilm. Thus, a value of Ag «0.45 eV has been obtainedfor gallium arsenide covered with a monatomic layerof cesium.[3SJ Activation of photocathodes by successivetreatment of the surface of the gallium arsenide withcesium and oxygen not only gives a smaller work func-tion, but it also gives rise to less bending of bands thana monolayer of cesium does. The value of Ag is0.25-0.3 eV for GaAs activated with a Cs-0 film.

Int3o:1, they studied the relation of the band bending tothe crystallographic plane of GaAs activated with aCs-0 film. It turned out that the (110) face commonlyused in single-crystal photocathodes is characterizedby stabilization of the Fermi level at the surfacecovered with a Cs-O film at the distance Ag =0.23 eV,and the (100) face has the close-lying value Ag = 0.28eV. The (lllA)-face having Ga atoms at the surfaceshows a considerably greater band bending Ag =0.86eV, while the (111B) face with As atoms at the surfacehas the smallest value Ag =0.1 eV.

In a specimen having a Z η impurity concentration-5X1018 cm"3, the width of the band-bending region atthe surface of the (111A) and (111B) faces amounts to155 and 51 A, respectively, while the escape probabil-ities of electrons are 0.21 and 0.49 (and of the orderof 0.3 for the (110) and (100) faces). The highest valueever measured of the integral sensitivity was obtainedwith a photocathode specimen prepared by activating a(111B) face of epitaxial GaAs with cesium and oxygen.It was 2060 μΑ/lumen (with an escape probabilityΡ = 0.58 and electron diffusion length L = 6 Mm).

There are two conceptions on the role of the acti-vating film on the surface of the GaAs, and in particular,

729 Sov. Phys.Usp., Vol. 16. No. 5, March-April 1974 N. A. Soboleva 729

on the explanation of the nature and mechanism of ac-tion of the coating, as obtained by repeated alternate(or simultaneous) deposition on the surface of cesiumand oxygen (up to 6-10 Cs-O layers): the heterojunc-tion model and the polarized-dipole model. The hetero-junction modelt».i5,i6,22,23,28,3o: t r e a t g t h e s u r f a c e l a y e r

as being a film of cesium oxide (assumed to be Cs2O),and ascribes to it the bulk properties of this material,which is a broad-band electronic semiconductor (gg= 2-2.2 eV) having a high donor concentration of ~1<?9

cm"3, a small electron affinity Sa=0-45 eV, and a lowwork function (which decreases with greater thicknessof the Cs2O film). According to the heterojunctiontheory, a potential barrier exists at the p-AHlBV-n-Cs2O boundary (a potential jump in the conductionband). The height of the barrier above the Fermi levelis determined by the relationship between the electronaffinities and the widths of the forbidden bands of thep- and η-type materials. In calculating the height of thisbarrier, one takes account of the fact that the CS2O filmis deposited on a GaAs surface already treated withcesium, i.e., on a surface whose work function has al-ready been reduced. An estimate of the minimum heightof the intermediate barrier at the GaAs(Cs)-Cs2O junc-tion with account for tunneling of electrons through itsupper part gives the value of 1.1 eV. Thus the inter-mediate barrier does not restrict the threshold forphotoemission from GaAs. When one uses materialshaving narrower bands, GaSb or ternary AEIBV com-pounds, the photoeffect threshold is determined by thegreatest of the three energies: the width gg of theforbidden band, the height SB of the intermediate bar-rier, or the work function e<p. Thus, in a study of ma-terials having gg<e<p, lowering the work function shiftedthe photoeffect threshold toward longer wavelengths untilthe work function became smaller than 83 · In agreementwith this model, the existence of a heterojunction barrierthat limits the photoemission threshold makes it impos-sible to prepare ΝΕΑ photocathodes having thresholdsλο>1.2-1.3 μΐη.

In agreement with the heterojunction model, one can-not shift the threshold to longer wavelengths solely byusing semiconductor materials having narrower for-bidden bands (and correspondingly having a smallerwork function so as to maintain the condition e<p < 8 g).One must take measures for simultaneous lowering ofthe height of the intermediate barrier, e.g., by specialtreatment of the surface of the material before activa-tion.: i o i : l The critical nature of the thickness of the ac-tivating coating arises from the fact that, on the onehand, increase in thickness is accompanied by loweringthe work function (owing to band bending in the Cs2Olayer), and on the other hand, the electrons that passthrough this film into the vacuum lose more energy.

Doubts on the correctness of the heterojunctionmodel were raised by a more exact estimate of thethickness of the activating film: as shown by Som-mer's1 2 4 1 measurements, which were carried out bychemical analysis instead of the previous estimatesfrom the fluxes of the introduced amounts of Cs and Οwith assumed high sticking coefficients, the amounts ofcesium and oxygen in the film proved to be practicallyindependent of the number of treatment cycles, butcorresponded approximately to one monolayer of cesium(~4xl014 cm"2) and one monolayer of CsaO (4.5X1014

cm"2). At this concentration, it is hardly appropriateto ascribe the bulk properties of the material to thecesium oxide.

In the polarized-dipole model, the greater reductionof the work function of the surface when oxygen is intro-duced into a cesium-treated specimen is attributed toformation of a double layer of charges and to the largerdipole moment of the combination GaAs-O-Cs comparedwith the combination GaAs-Cs.C41'42] This model is con-firmed by experiments on an electronically-stimulateddesorption of cesium from substrates not subjected andsubjected to oxygen treatment.1''433 Recent experimentson infrared photocathodes based on ternary A ^ B ^ com-pounds158"6 have shown that the assumed limitation ofthe threshold of the photoeffect by the height of the inter-mediate barrier of a heterojunction (-1.1 eV, indepen-dently of the width of the forbidden band of the substratematerial) is apparently not confirmed: it was possible touse InAsP and InGaAs with a monolayer Cs-0 coatingto prepare photocathodes having a photoemissionthreshold hiO = 0.9 eV (λο= 1.4 μηι), which was limitedonly by the width of the forbidden band of the material.The lowered photoemission quantum yield observed hereis attributed to a decrease in the escape probability Ρof the electrons, which apparently involves a decrease inthe absolute value of the negative electron affinity.

While there is as yet no exact quantitative theory ofthe action of the activating coating, experiment showsthat the nature of the surface structure that arises inactivation and the potential diagram of the surface re-gion apparently depend not only on the relationship be-tween the energy constants of the substrate and coatingmaterials, but also on the state of the surface of thesemiconductor, and on the method of cleaning andtreating it.

The state of the surface, the stability of this state, andthe effect of various contaminants on the photoemissionproperties of ΝΕΑ cathodes have been studied withvarious modern methods of surface analysis: mass spec-trometry,U 4 1 Auger spectroscopy,U5>46] the method ofelectron-stimulated desorption, and study of electron-reflection characteristics. An infinitesimal surfacecontamination by carbon, which is not removed in theusual methods of thermal cleaning, has been shown toexert a harmful influence on the sensitivity of photo-cathodes (this implies that oil-less pumping means areneeded). It has been shown that instability of photo-emission (lowering of the quantum yield in the near-threshold region of the spectrum) involves instabilityof the surface activating film, due to desorption ofcesium in a continuously pumped space or adsorptionof residual gases in sealed devices. Here the escapeprobability Ρ of the electrons decreases; the bulk prop-erties (the diffusion length of electrons) hardly vary.[ 3 i :

We must note that, although a Cs-O activating coat-ing is stabler than a monolayer of Cs, introduction ofoxygen increases the desorption cross section of thecesium under electron bombardment. This implies thatit seems expedient in preparing ΝΕΑ secondary-elec-tron emitters to limit the activation of the surface tocesium alone, rather than Cs-O.

Films of CsF,CsOH,BaO, etc.,have also been used asactivating coatings, instead of cs-O." 1 " 1 3 ' 2 1 ' 2 8 ' 5 3 ' 5 6 ' 6 5 ' 6 9 3

In line with the fact that the activating film on thesurface of ΝΕΑ emitters must have a quite definite op-timum thickness, a need has arisen for new, high-puritycesium sources with flux intensities that can be regu-lated and monitored. Most research centers have usedfor this purpose either molecular cesium sources witha guided flux, e.g., "channel" sources in the form of a

730 Sov. Phys. Usp., Vol. 16, No. 5, March-April 1974 N. A. Soboleva 730

nickel tube filled with a mixture of cesium chromateand Si, or ion sources based on cesium aluminosilicates,impregnated with cesium carbonate, aluminum, or usingsurface ionization of Cs on a hot tungsten tube. The ad-vantage of the ion sources is that they permit one tomonitor the degree of covering of the surface from thesize of the ion current.U 7"4 9 ] To introduce a controllableamount of oxygen, one uses either flow regulators, whichuse diffusion of oxygen through a heated thin-walledsilver tube, or thermal release of oxygen from higheroxides (MnCb, BaO2, etc.).

Practical realization of the principle of building elec-tron emitters having a negative electron affinity andpreparation of such photo- and secondary emitters insealed devices became possible only because of a highlevel of technology, because of development of methodsof preparing high-quality semiconductor materials withperfect crystal structure, development of methods ofcleaning their surfaces and optimizing the means of ac-tivation, and also because of invention of commercialsuperhigh vacuum apparatus (one needs a vacuum noworse than 10~9 Torr to prepare ΝΕΑ emitters).

Table I gives the fundamental information on experi-mental photocathodes based on gallium arsenide pre-pared by cleaving single crystals in a superhighvacuum (Sec. a of Table I) and on commercial typesof photocathodes prepared by using polycrystalline orepitaxial films of this material and cleaning the sur-face by brief radiation heating to a temperature closeto the dissociation point of this compound (-650° C).As we see, optimization of the method of preparing thematerial, the doping level, and the surface activationhas permitted gaining a mean level of integral sensi-tivity of gallium-arsenide photocathodes of -1000μΑ/lumen (with a thermal emission of ~10"16 A/cm2),even in sealed devices, with twice as high a sensitivityin experimental apparatus. Figure 4 shows the spectralcharacteristics of the first GaAs-Cs photocathode ofScheer and van Laar in 1965 (on a cleaved single crys-tal) and of two modern gallium-arsenide photocathodes,one being a cleaved GaAs(Zn)-Cs-O photocathode having

a sensitivity of 1700 μΑ/lumen, and the other an epi-taxial GaAs(Zn)-Cs-O photocathode (with a [l l lB]surface) with a sensitivity of 1837 μΑ/lumen.C383 Thedevelopment of new ideas on the factors that affectthe photoemission efficiency has facilitated in recentyears a considerable refinement of ordinary photo-emission materials. For comparison, Fig. 8 shows thecharacteristics of a standard (S-20) and a new modifi-cation (ERMA-ni) of a multialkali photocathode. Figure5 gives the spectral characteristics of two types ofNEA-photocathodes used in commercial types of photo-

FIG. 4. Spectral characteristicsof photocathodes. J -GaAs-Cs[4];2—GaAs-Cs-O with a sensitivity of1700 μΑ/lumen [ 1 0 1 ] ; 3-GaAs-Cs-O(111Β face) with a sensitivity of1837 μΑ/lument30]. Dotted lines-the spectral characteristics ofordinary photocathodes: the S-20multialkali cathode and the ERMA-III cathode with increased redsensitivity.

τη/

11/ί11ι·III

: =

^ * "

1

11

3 =

— — -

• s-

n r —

A-m

20

τ

ζ

ERMA

10

Si, mA/watt

2,0

FIG. 5. Spectral characteristics ofcommercial NEA-photocathodes usedin photomultipliers by the RCA firm.The dotted line shows the character-istics of standard photocathodes [101 ] •

F

a

'/1

-

AsΡ

\

S-10BHg-O-Cs

Λ

\

\

CLAS •

EHM

1

1

J0.4 0.S

λ, urn

TABLE I. Fundamental Properties of Photoemitters Based on GaAs

Material.

a) Singlecrystal(110)Singlecrystal(110)

Single cry-stal (110)Singlecrystal(110)Singlecrystal

,(110)Single cry-stal (100)

Singlecrystal

f (111B)b) Poly-crystal(evap.)Vaporepitaxy

Epitaxy:Liquidepitaxy

Epitaxy

Dopant,cm"3

Zn4-101·

Zn2-6-10»

Zn 10»

Zn4-10»

Z D

2,6-1018-5-10»

Ζ η 1018Si

Zn5-1018—

MO»Zn

5-1018Zn

5-10»

Zn1.5-10»

ZnGe

5-101'—2-1018

Zn

Cleaningof surface

Cleavage inultrahighvacuumCleavage inultrahighvacuumHeating to650°CCleavage

»

1 Heating

Ionbombard-mentHeatingHeating

Heating

Activation

Cs

C s — 0

Cs — 0

Cs+(O+Cs)«

Cs — 0

CsF, CsCsF

C s — 0

C s — 0

Cs

C s — 0

Cs — 0Cs — 0

Cs — 0

Integralsensitivity,μΑ/lumen

500

500

600

1000

1100

12401700

2060

220

423

5001100

1000

Ρ

Γ,μπι

—

0.11

0,37

0.45

0,5

0.55-0.6

0.6

—

χ, Α

—

0,47

0.6

0.5

1.2-3

6

5 - 7

—

L

Γ,JUIil

—

1.45

1,2

1.38

—

X.A

—

1000

300

1200

_

—

I

Refer*ence

4

8

Π

b

20

212 β

101

-30

fll

β 5

24

27

33

731 Sov. Phys.-Usp., Vol. 16, No. 5, March-April 1974 N. A. Soboleva 731

FIG. 6. Spectral characteristics of twophotocathodes based on GaAsSb[57'104].Dotted line—silicon NEA-photo-cathode[5 9]. The characteristic of aGaAs—Cs—Ο photocathode is given forcomparison.

multipliers by the RCA firm. As we see, the ΝΕΑphotocathode has a high and uniform efficiency up to awavelength λ~850 nm, where the spectral sensitivityof the GaAs cathode exceeds 100 mA/W, while it is15-20 mA/W in the best multialkali photocathodes. Itis inferior to the multialkali photocathode only in thespectral range X>900nm.

In order to prepare infrared-sensitive photocathodes,materials having narrower bands are used, mainlyternary AHIBV compounds, whose widths of forbiddenbands can be smoothly varied by varying the ratio ofcomponents in the solid solution. : 5 0 ' 5 8 ' 6 0 - 6 2 ] It wasshown possible in 1970 to achieve a negative electronaffinity in silicon.C59>84] Although silicon is distinguishedby its very large minority-carrier diffusion length, itcould hardly be classified as a promising material forpreparing photocathodes: having indirect bands, siliconhas a low optical absorption near the threshold. Thisleads to a smoother trend of the spectral characteristicin the near-threshold region. Silicon can be success-fully used to prepare other types of ΝΕΑ emitters, e.g.,secondary or cold cathodes. As we have stated above,the shift of spectral characteristics to longer wave-lengths as the width of the forbidden band of AUIBV de-creases is accompanied by impairment of form and de-crease in quantum yield .2) It has been possible to shiftthe photoemission threshold without substantiallylowering the quantum yield by using films of GaAsxSbj-χof variable composition, having a forbidden band thatnarrows smoothly in the direction toward the surface.Owing to the slop^ of the lower boundary of the con-duction band, such films contain an internal field thatfacilitates transport of excited electrons to the sur-face.t 5 7 ] Figure 6 shows the spectral characteristic ofthis cathode, as well as the characteristic of a homo-geneous GaAsSb cathode having a forbidden-band widthg g = l . l eV.1·104-1 This same graph shows the spectralcharacteristic of an ΝΕΑ photocathode made of siliconhaving the same width of forbidden band.C59]

The best results in making efficient photocathodes forthe infrared have been obtained with the ternary com-pounds InAsi-χΡχ and InxGai-XAs. Figure 7 shows thespectral characteristics of two photocathodes based onInASi-χΡχ (x = 0.6 and 0.85) according to C58], togetherwith the envelope curve that shows the maximum sensi-tivity of photocathodes of this composition as optimizedby varying χ and the thickness of the activating coatingfor operation in this spectral region.[6o:i For compari-son, the dotted lines show the spectral characteristicsof a multialkali (S-20) and a silver-oxygen-cesium (S-l)photocathode. Figure 8 shows the spectral characteris-tics of several photocathodes based on InxGai-xAs. The

FIG. 7. Spectral character-istics of photocathodes based onInAsP[5 8'6 0].

-

sSi

—

)M

Γ\1

"-20

1,06

&

In

1•ail

Λ

0.85

\

0.6 0.8 1.0 ί.2 ΙΑ

10

FIG. 8. Spectral character-istics of photochathodes based onInGaAs[6 1'1 0 4]. The dotted lineshows the characteristic of a semi-transparent photocathode of this 10composition!!2].

10

10 0.6 0,1 i.o a

TABLE II. Fundamental Data on Infrared Photocathodes

Material

InPIn^Gai.iAsχ — 0.11i = 0.16i = 0,25J = 0.3i = 0-0,221 = 0,14-0.31InASxPj.-c1 = 0.151 = 0.15—0,4i = 0-0.26InAsP

GaAsxSbi_x(with variable x)Si

Activationj

Cs —0CsOH

Cs-0Cs — FCs — 0To me» »» »

» »

» »» »

Integralsensitivity,μΑ/lumen

450245

130

__

8001126

1300700

—

%g,Λνο, eV

1.241.16

_

1.00,95

1.4—1.08To 0.9

1.1

To 0.9

1.05 (in asealeddevice)1.11.06

1.1

Quantum yield Υ ;

at 1.06 μην

-

5-10-3

4-10-*2-10-28-10"3

4-10-32-10-2

.

2-10-3

at 0.94 μηι

—

10-2

5.5-10-2

—

Refer-ence

50

5 1

98

53

6 1

«2

6 1

58

58

60

63

eo, ιοί57

59

thin lines refer to the cathodes in commercial sealeddevices.11043

The visible lack of a limitation of the photoemissionthreshold by an intermediate barrier (but with limita-tion only by the width of the forbidden band of the ma-terial) permits us to expect that we can shift the photo-emission threshold as far as wavelengths of -1.8 μπι byfurther improvement of the quality of the used materialsand means of activating them (in line with the actuallyattainable minimum work function ~0.7 eV). Table IIgives the fundamental information on ΝΕΑ photocathodessensitive to the infrared.

The first reports of commercial production of photo-multipliers having gallium arsenide (and gallium ar-senide-phosphide for the visible spectrum) photocathodesappeared in 1969. The first devices with photocathodessensitive to the infrared were already being produced in1970, and they were designed for detecting infrared ra-

732 Sov. Phys.-Usp., Vol. 16, No. 5, March-April 1974 N. A. Soboleva 732

diation of light emitting diodes and lasers. For opera-tion with light emitting diodes and injection lasers basedon gallium arsenide (λ= 930-950 nm), one uses mainlyphotocathodes based on InxGai-xAs (0.05<x<0.10). Ithas a spectral sensitivity in this spectral region of-30-40 mA/w and a thermal emission that does notexceed 10"16 A/cm2. At this wavelength, ordinaryphotocathodes, e.g., the S-20, have a spectral sensi-tivity of ~0.1 niA/W, and S-l photocathodes have ~l-2mA/W (and the latter have a thermal emission at leastthree orders of magnitude greater!). Photocathodes ofthe same composition optimized for operation in the re-gion of Nd laser emission (1.06 Mm) are characterizedby a larger percentage of In (0.15 <x< 0.25). Their spec-tral sensitivity at 1.06 Mm is as great as 15-17 mA/W(aquantum yield of ~2%). That is, it exceeds by a factor ofseveral tens the sensitivity of the standard S-l cathode,while the thermal emission amounts to only ~1015 A/cm2.

On the practical level, the development of technolog-ically feasible methods of preparing ΝΕΑ photocathodesin the form of films is of greatest interest, and es-pecially thin semitransparent films on transparent sub-strates for use as cathodes illuminated from the back("transmission mode").

A series of studies has calculated^9"713 the optimumthickness of the photosensitive layer for getting themaximum sensitivity in illumination from the back andthe relation of this sensitivity to the quality of the ma-terial (the diffusion length for electrons), to the defectdensity at the cathode-substrate boundary (the rate ofsurface recombination), and to the optical propertiesof the system (reflection at the cathode-vacuum andcathode-substrate boundaries). Analysis of the opera-tion of a semitransparent GaAs photocathode shows thatnear the threshold of the photoeffect (up to hv= 1.45-1.75 eV) one can get a photoemission quantum yield thateven exceeds the corresponding quantity in direct il-lumination with a thickness of photosensitive film notexceeding the diffusion length for electrons. However,a number of conditions must be satisfied (absence oflattice-mismatch defects at the cathode-substrate bound-ary, and optical coating of the illuminated surface). Thespectral dependence of the quantum yield in illuminationfrom the back has a maximum at a definite value of theproduct ah that depends on the thickness of the film (ais the optical absorption coefficient of the material).

In order to prepare semitransparent photocathodes,one needs high-quality materials not inferior in theirbulk properties to the single crystals used in preparinggood opaque cathodes. We can cite the following calcu-lated relation.

A GaAs photocathode having L = 1.6 Mm and Ρ = 0.37has at hf= 1.5 eV a direct-illumination quantum yieldΥ = 0.139. A semitransparent photocathode made of thesame material with a film thickness of 1.2-1.6 Mm canhave a quantum yield -0.144 in the absence of surfacerecombination without coating, or 0.216 when coated.However, if the diffusion length proves to be an orderof magnitude shorter (0.16 Mm), the quantum yield inbackward illumination decreases by more than an orderof magnitude, while it requires a considerable decreasein film thickness to optimize it. Owing to the small op-tical absorption coefficient pear the photoemissionthreshold (a ~ 1 Mm"1 for GaAs), the diffusion lengthfor electrons in semitransparent films must be greaterthan 1 Mm. Preparation of thin tilms (of the order of

1 Mm) that maintain a high quality of material presentsgreat technical difficulties. One can get very thin filmsby vacuum sputtering (the three-temperature method) *" ,but they are polycrystalline with a fine-grained struc-ture and have a diffusion length of only several hundredAngstroms.

Films having L s 4000 A have been prepared by amodified cathode-sputtering method.1-7 One gets thebest quality of films by preparing them by the methodof vapor-phase or liquid epitaxy, but starting only with acertain, not too small thickness. Liquid-phase epitaxypermits one to monitor exactly the stoichiometry, thedegree of doping, and the structure of the films duringthe growth process/6 6 3 but it requires a subsequentcleaning of the surface by etching. Vapor-phase epitaxyis the most practically-developed method, which doesnot require subsequent etching.

The need for a low defect density at the substrate-cathode boundary requires a careful choice of the sub-strate material. People have tried to use as substratesglass, fused or single-crystal sapphire, and silicon,which is transparent in thin films in the long-wave-length region. In order to maintain good structurein the transverse direction (normal to the surface ofthe film), polycrystalline films were prepared withgrain sizes equal to the film thickness (0.6-0.8 Mm).[71]

Even here, however, the sensitivity of the photocathodesin transmission remained more than an order of magni-tude below that in direct illumination. It seems mostpromising to prepare epitaxial films on crystallinesubstrates having close-lying lattice parameters, i.e.,on AHIBV compounds having wider forbidden bands.Semitransparent photocathodes have been describedthat are made of GaAsSb (gg= 1.15 eV) on a substratemade of chromium-doped gallium arsenide, which istransparent to 1.4 eV, and of GaAs on an AlGaAssubstrate.t75'93 : l The best results have been obtainedwith photocathodes made of InGaAs on substrates ofGaAs andGaP.t62] We should note that substrates ofthis type serve as optical filters that restrict the sensi-tivity of the photocathodes in the short-wave region ofthe spectrum.

As the technology of preparing materials havinggreat diffusion lengths for electrons is perfected,people will be able to solve the problem of makingsemitransparent photocathodes without substrates inthe form of mechanically strong free films (plates)several microns thick.

Table III gives the fundamental information on ex-perimental semitransparent photocathodes based onAHIBV, and Fig. 9 shows the experimentally obtainedspectral characteristics of these cathodes made ofGaAs and of two types of ternary compounds based onGaAs.

Use of materials having negative electron affinityas secondary-electron emitters is facilitated by thefact that one can use for this purpose semiconductorshaving wider forbidden bands, and the conditions fornegative affinity are satisfied at higher values of thework function than for photoemission. Consequently,secondary emitters are less sensitive to poorer vacuum.This made it possible to produce as early as 1968 pho-tomultipliers having dynodes (or with a single firstdynode) made of gallium phosphide (gg = 2.3 eV) acti-vated with a cesium film. The great escape depth of

733 Sov. Phys.-Usp., Vol. 16, No. 5, March-April 1974 N. A. Soboleva 733

TABLE III. Fundamental Information on SemitransparentPhotocathodes.

Material

GaAs(Zn)

GaAsSb

GaAs

GaAs

GaAs

GaAs

GaAs

GaAs

GaAs(Ge)InGaAs

Depositionmethod

VaporepitaxyLiquid 1

epitaxy •VapotepitaxyEvap. Ί

Vaporepitaxy •Evap. ,

CathodesputteringLiquidepitaxy

Liquidepitaxy

Filmthickness,

izm

2.5

2.0

2.5-0,85

2.0

0.6-0.8

0.35-1.0!

10-130.1-1.0

Substrate

Sapphire

GaAs(Cr)

Sapphire

Glass,sapphire,'quartz {Sapphire

Glass,sapphire,quartz

Singlecrystal. sapplure:..

Silicon '

AlGaAsGaAs,GaP

Integral t

sensitivity, ;μΑ/lumen '

Re-flec-tion

100

—

-

-

-

-

.

—

1000—

Trans-mis-sion

1

—

-

10

70

-

—

Quantumyield, Υ ;

Re-flec-tion

8-10-·2-10-3

—

-

-

—

3-10-·

2-10-'10-»

—

.3-10"3

Trans-mis-sion

10-310-310-2

6-10-2

5-10-3

10-2

3 -10-»

2-10-·2-10-3

—

3-10-3

eV

1,41.51,37

1.45

3,0

1.7

1.5

1.53.0

1.25

Refer-ence

68

89

70

G7

7 1

08

73

74

•3

6 2

g if

10 -•

l v — ^

A%

j \ \12

FIG. 9. Spectral character-istics of semitransparent photo-cathodes. 1 -InGaAs onGaAs[2];2-GaAsSbonGaAs[ 6 »];3-GaAs(2^mthick) on sapphire!70]. Thedotted lines are the character-istics of the correspondingcathodes in direct illumination.

US 0.7 0.8 fti 1.0 Ο 1.2λ , Mm

secondary electrons in ΝΕΑ emitters gives rise to ex-tremely high values of the secondary-emission coeffi-cient (SEC) in the high-energy region of primary elec-trons. Thus, the calculated SEC (σ) of GaP-As at themaximum is am~25O at Up~20 kV.C76] At Up = 2.5 kV,σ= 130 has actually been obtained, and σ= 30-50 atUp=600 v.t77"83: l A secondary-emission coefficient-400 has been obtained in epitaxial GaAs—Cs-0 atUp = 20 kV.:80] There is a report of a calculated valueof CTm~ 1000 for this material.[ 8 1 ] The lack of any re-striction by optical properties permits one to use morewidely as secondary emitters materials having a greatdiffusion length, e.g., silicon, by preparing from themtransmission secondary emitters in the form of freethin plates (films) without supporting grids. Figure 10shows the emission characteristics of silicon activatedwith Cs-0 upon electron bombardment of a plate 4-5Mm thick on the front ("reflection") and back ("trans-mission") sides.[ 8 4 ] It also shows the characteristicsin reflection of the emitters GaP-Cs, GaP-Cs-O, andGaAs-Cs-O, and for comparison, one of the materials(CsaSb) commonly used in photomultipliers. Appar-ently, the use of silicon diodes in photomultipliers willbe hindered by their high thermal emission, which is asmuch as ~10'10 A/cm2 at room temperature.C l 0 6 ] Thereis a description of a transmission secondary emitterbased on GaP in the form of a film ~0.2 μπι thick, epi-taxially grown on GaAs with the substrate subsequentlyetched away. At Un= 2 kV, the SEC of such emitters isas great as -100. In thin plates of GaAs (3-5 μΐη),people have obtained at Up= 10 kV an SEC ~115 in reflec-tion and up to 30 in transmission.[8o:l

FIG. 10. Emission charac- *teristics of secondary emitters.1 —Si—Cs—Ο is reflection,2—in transmission!84], andGaP-Cs; above-emissioncharacteristics in the regionof low primary voltages forGaP-Cs[76>85], GaP-Cs-O[8 3]andGaAs-Cs-O["];thesecondary-emission character-istic of Cs3 Sb is given forcomparison.

400

4 1 U J

2

I

// r• ,

Q0 41

/

>

W Λ

/

/

&

U

Γ

GaP-Cs-C

GoA6aP

A

• v

/

s-Cs-0

/

/

//

•r

Λ

fin

Si

p-r.s

2 ί 6 8 10 II « № 16 20

Currently the American industry produces up to 40types of photomultipliers having ΝΕΑ emitters. Use ofGaAs-Cs-0 photocathodes with a mean integral sensi-tivity of ~1000 μΑ/lumen and thermal emission ~10"16

A/cm2 in a photomultiplier permits a considerableimprovement of their threshold characteristics, and itmakes of these devices unique photodetectors that candetect single photons.[86) Use of first dynodes made ofgallium phosphide with an amplification of ~50 insteadof 3-5 substantially improves the statistics of theemission processes. This makes it possible to reducethe noise factor introduced by the dynode system from0.25 for the best ordinary photomultipliers to 0.04. Thusthe photomultiplier becomes an ideal device in whichamplification is not accompanied by an increase innoise, and the sensitivity threshold is determined solelyby the statistical fluctuations of the thermal emissionof the photocathode. With slight cooling to -20° C, thedark current in these photomultipliers amounts to ~2pulses per second.

The high amplification of the first dynode sharplyimproves the amplitude resolution of the photomultiplier.This is especially important in detecting very weakscintillations from low-energy radiation. The photo-multipliers have a narrow electron distribution, with ahalf-width for a single-electron pulse of ~40%, and theypermit one to distinguish clearly up to six electron peaks.

In line with the narrowing of the energy spectrumof the photo- and secondary electrons, the time resolu-tion of the photomultipliers is also considerably im-proved. There is a report of an improvement of thetime resolution of single-photon pulses by 40% as com-pared with the photomultipliers having copper-berylliumdynodes.[87] The RCA-C31024 photomultiplier having fivedynodes made of GaP-Cs is characterized by a leadingfront of single-electron pulses of ~0.8 nsec and a pulseduration of ~1.5 nsec at an amplification of 108.C88i Themost promising line of further improvement of thephotomultiplier seems to be that of using transmissiondynodes, which permit considerable improvement of thetime resolution. CalculationsCl06:l show that when oneuses one transmission dynode made of GaAs with anamplification of -100, one can get a duration of the lead-ing front of -37 nsec and a pulse half-width of -135 nsec.Measurement of the service life of devices using GaP-Csdynodes has shown that they are at least as stable as de-vices with the ordinary alloy dynodes.

The high secondary-emission coefficients render thenew materials extremely promising also in use as tar-gets for identification of particles at relativistic ener-

[81]

The principle of negative electron affinity is fruitful,not only as applied to photo- and secondary electron

734 Sov. Phys.-Usp., Vol. 16, No. 5. March-April 1974 N. A. Soboleva 734

emitters, but also to other types of unheated electronemitters.

The possibility of using forward-biased surface-barrier and p-n junctions as injection cold cathodes hasbeen studied intensively over the past 5-10 years. Dis-covery of the phenomenon of negative electron affinityhas facilitated a sharp improvement in the parametersof this type of electron emitter as well. The principle ofaction of injection cold cathodes with negative electronaffinity is that, when a forward bias is applied to a p-njunction having an outer p-layer, electrons are injectedinto the conduction band of the latter. They diffuse tothe surface and escape into the vacuum if the thicknessof the p-region does not exceed their diffusion lengthand there is no potential barrier at the surface. Theelectron-injection coefficient a is designated as theratio of the number of electrons injected into the p-regionto the total number of carriers (electrons and holes)that constitute the direct current across the junction.The fraction β of the injected electrons that arrive atthe surface of the p-layer is determined by the relation-ship between the thickness of the p-layer and the diffu-sion length of the electrons. The escape probability Ρ ofthese electrons into the vacuum depends on the samecircumstances as in photo- and secondary emission fromsimilar structures. The efficiency of a cold injectioncathode, as defined as the ratio of the emission currentto the current across the junction, is equal to theproduct of these quantities:

/,//„_„ = α-β·Ρ. (10)

The efficiency depends on the material that the cathodeis made of, on the technical possibilities of producingan outer p-layer of small enough thickness (in practice,several micrometers), and on the treatment of thesurface. Considerable difficulties are involved withthe need to create current-carrying contacts that donot disturb the structure of the thin junction.

A cold injection cathode based on an epitaxial p-njunction in GaAs with a surface activated with cesiumand oxygen was first made in 1969.C89:| Its efficiency waslow, mainly because of an unfortunate geometry: theemitting surface was a lateral section of the p-n junc-tion. Subsequently, a flat, thinly deposited p-n junc-tion in silicon was used for this purpose. A note i9o:

reports on getting an efficiency of cold emission of~2%, and a stable emission current of ~7.3 mA (at acurrent density of ~1 A/cm2).

A number of foreign companies are vigorously in-volved in developing commercial injection cathodes.There are reports on preparation for sale of single andmatrix p-n cathodes made of gallium phosphide, andon obtaining current densities up to several amperesper cm2 in a steady-state mode, and more than 100A/cm2 in a pulsed mode.C9i:1

At the same time, a second variant of the negative-electron-affinity cold cathode has been developed: theoptoelectronic cathode, in which optical excitation ofelectrons in the outer p-layer is used instead of in-jection into it, as the radiation of a light diode is ab-sorbed in it. The optoelectronic cathode is a combina-tion of an electrically controlled light source with asemitransparent photocathode. The efficiency of anoptoelectronic cathode, i.e., the ratio of the emissioncurrent to the light-diode current, is determined by theproduct of the quantum efficiency of the light diode, the

transfer coefficient of the radiation to the light-sensi-tive layer, the absorption of radiation in it, and thequantum yield of photoemission in the spectral regionof emission of the light diode.

The appearance of gallium arsenide photocathodeshaving negative electron affinity immediately stimulatedthe invention of optoelectronic cold cathodes in which asemitransparent GaAs-Cs-O photoemitter had beendirectly deposited (with direct optical contact) on theemitting p-region of the light diode. The emission ofthe light diode occurs at a quantum energy somewhatsmaller than the width of the forbidden band of the ma-terial. Hence, one chooses as the light-diode materialan A ^ B V semiconductor having a width of forbiddenband slightly exceeding the width of the forbidden bandof GaAs, e.g., AlGaAs having low Al content (at thesame time, this makes it possible to grow high-qualityepitaxial GaAs films on its surface). With a forwardbias on the p-n junction in the AlGaAs, electrons areinjected into the p-region, where they radiatively re-combine. The radiation (ην =1.4-1.6 eV) is absorbedin the p+ layer of the GaAs grown on the p-region ofthe light diode. The generated photoelectrons diffuseto the surface, and they escape into the vacuum withoutsubstantial losses if the p-GaAs layer has the optimumthickness for a semitransparent photocathode.

The efficiency of the first cold cathodes of this typeamounted to ~10~3. This was explained mainly by in-sufficient sensitivity of the thin (~1 Mm) semitransparentphotocathode.C92] The design and means of preparingoptoelectronic cathodes have subsequently been refined:a photosensitive layer made of high-quality galliumarsenide doped with germanium was deposited by liquid-phase epitaxy, and it had a thickness of 10-13 μηι,which permitted more complete absorption of the radi-ation. The surface of the GaAs was activated by cesiumand oxygen until maximum photosensitivity had beenattained, being 700-1000 MA/lumen.C93:l Figure 11 showsthe structures of an injection cathode made of Si andof an optoelectronic cold cathode made of GaAs havinga working surface of ~10"2 cm2. The ohmic contactwith the n-region was made with In, and that with thep-region was made with a pressure point contact. Theefficiency of the optoelectronic cathode was as muchas 1.6-2%, and the pulse current density as much as3 A/cm2. An approximate estimate of the theoreticallimit of the efficiency when the thickness of the photo-sensitive layer is optimized, and with an internalquantum efficiency of the light diode of ~50%, givesa value of ~4%, if the escape probability of electronsis-0.15.

As a number of technological problems are solved,mainly involving the technique of creating contacts, andwith further increase in the emission current density,cold cathodes of both types may successfully replace

FIG. 11. Design of an injectioncold cathode based on silicon [90]and of an optoelectronic coldcathode!9 3].

735 Sov. Phys.-Usp., Vol. 16, No. 5, March-April 1974 N. A. Soboleva 735

FIG. 12. Diagram of the photo-emission of a reverse-biased p-njuntion through the thin outern-region[94].

thermoelectronic cathodes in a number of electronicdevices. These emitters are of especial interest forelectron-ray devices, in particular, for creating many-beam projectors with independent control of the beams.

In closing the review, I might also mention a numberof as yet not achieved projects to invent efficient photo-cathodes having emission stimulated by an electric fieldin which conditions of negative electron affinity arealso used. The idea is known of using the strong elec-tric field that arises in the depleted region of a reverse-biased p-n junction for heating of electrons that havebeen optically excited in the ρ region in order tofacilitate their escape through the surface potentialbarrier. Applying a reverse bias to a p-n junction hav-ing an outer n-region lowers the vacuum level withrespect to the bottom of the conduction band of the ρregion, and it can give rise to a pattern analogous tothe conditions of negative electron affinity. As we seein Fig. 12, the electron affinity for electrons excitedin the p-region is defined by the expression

g. = «φ - %, - eU. (11)

In order to produce such photocathodes, one must createstructures having very sharp p-n junctions, with athickness of the n-region comparable with the phonon-scattering length of hot electrons. The problem is dis-cussed in of choosing optimal materials for buildingthis type of photoemitters that are sensitive fartherinto the infrared than the existing ΝΕΑ photocathodesare. The fundamental criterion is a small width offorbidden band, a large phonon scattering length of hotelectrons, and the possibility of creating a strongenough field at the junction without having a tunnel cur-rent and avalanche breakdown. By using an approxi-mate quantitative estimate of the parameters of thetransitions, it has been shown possible to create p-nphotoemitters based on InAs. This requires creatingan η-type surface layer with a donor concentration of~1015 cm"3 and a thickness of ~0.6-0.8 μπι. One mustcreate a very thin n+-layer activated with Cs and Ο toreduce the work function in order to permit contactwith the surface of the n-region. The possibility ofmaking photoemitters based on materials having evennarrower bands, InSb and HgCdTe, has been discussed.

Patent descriptions exist on photoemitters of thistype based on a Ge-GaP heterojunction and silicon.[95'96]

In order to create a sharp fine-scale p-n junction on thesurface of a p-Si plate, it has been proposed to use themethod of ion implantation by 5-10 keV cesium ions,which become implanted to a depth of 50-200 A. Thesurface is activated with slow Cs ions to reduce thework function. An annular diffusional η-region pro-vides for contact with the outer layer of the cathode.

The principle of action of an infrared photocathodehaving two heteroj unctions (with a transistor structure)has been discussed in t 9 7 3 for the system p-Ge-e-ZnSe-

FIG. 13. Energy diagram of anexternally-biased three-layer cathodestructure[97].

v-ZnSe

p+-GaAs · (Cs-O) having a surface activated to the pointof negative electron affinity. Figure 13 shows the en-ergy diagram of this structure with an applied externalbias. In essence, this system is a combination of anilluminated, reversed-biased junction and an outeremitter junction through which electrons are injectedfrom the basis into the p-type gallium arsenide. Theouter layer is an ΝΕΑ photocathode in which optical ex-citation has been replaced by injection. The electronsthat are excited by infrared radiation in the intrinsicband of the germanium (upon illumination from the sub-strate side) diffuse through the thin base (ZnSe), areinjected into the GaAs, and escape into the vacuum,provided that the thickness of the GaAs layer does notexceed their diffusion length. The components of thethree-layer structure were chosen with account takenof the idea that no potential jump (barrier) should arisein the conduction band at the boundaries of the twoheterojunctions. The large width of the forbidden bandof the base material (ZnSe) hinders the flow of a largehole current through the emitter junction.

There is as yet no information on the practical re-alization of photoemitters having an externally biasedjunction. To prepare them involves the great difficultyof producing structures having an extremely thin outerjunction region (a fraction of a micron) with sufficientconductivity to ensure contact.

More than 100 original studies and several reviews148"106 ]

already have been concerned with describing the newclass of efficient electron emitters and solving theproblems that arise in preparing them. The rapid re-finement of the parameters of the new emitters revealsthe great interest of industry in producing devices hav-ing emitters that permit one to take a qualitative jumpin an entire series of very important properties of thesedevices. The scientific value of the idea of producingand using negative electron affinity is unquestionable:the creation of a broad class of various emitters withnegative electron affinity is one of the very greatachievements of physical electronics throughout itshistory of existence.

''The electron affinity S a of a semiconductor is taken to be the energyspacing from the bottom of the conduction band to the vacuum level,which determines the height of the potential barrier for conductionelectrons.

2)One of the reasons for this can be simply the increase in density ofcrystal-lattice defects as the percent composition of the narrow-bandcomponent of the compound increases.

'W. E. Spicer, Phys. Rev. 112, 114 (1958); J . Appl. Phys.31, 2077 (1960).

2W. E. Spicer, A. H. Sommer, Photoelectronic Ma-terials and Devices. Ed. S. Larach, 1965.

3A. H. Sommer, Photoemissive Materials, 1968.4 J. J. Scheer and J. van Laar, Solid State Comm. 3, 189(1965); West German Patent cl. 21d 29/20, No. 1256808,Dec. 2, 1954.

5 J. W. James, J. L. Moll, and W. E. Spicer, Gallium

736 Sov. Phys.-Usp., Vol. 16, No. 5, March-April 1974 N. A. Soboleva 736

Arsenide. Proc. 2nd Intern. Symposium, 1968.6L. W. James, J. L. Moll, Phys. Rev. 183, 749 (1969).7R. C. Eden, J . L. Moll, and W. E. Spicer, Phys. Rev.Lett. 18, 597 (1967).

8A. A. Turnbull and G. B. Evans, J. Phys. Dl, 155 (1968);Electronics 40, 234 (1967); British Patent (H 01 j 9/l2)No. 1200899, Aug. 30, 1967.

9T. E. Fisher, Surface Sci. 10, 474 (1968).10H. Sonnenberg and R. T. McKenzie, IEEE Trans.

ED-15, 419 (1968).n P . G. Borzyak, Yu. G. Zav'yalov, G. A. Katrich, and

O. G. Sarbei, Ukr. Fiz. Zh. 14, 403 (1969).12M. Hagino and R. Nishida, Japanese J. Appl. Phys.

8, 123 (1969).13H. Sonnenberg, Japan J. Appl. Phys. 8, 806 (1969).14H. Sonnenberg, J. Appl. Phys. 40, 3414 (1969).15H. Sonnenberg, Appl. Phys. Lett. 14, 289 (1969).16H. Sonnenberg, Bull. Am. Phys. Soc. 14, 791 (1969);

Appl. Phys. Lett. 21, 103, 278 (1972).17Y.-Z. Liu, J. L. Moll, and W. E. Spicer, Appl. Phys.

Lett. 14, 275 (1969).18R. L. Bell, Sol. St. Electron. 12, 475 (1969).19R. L. Bell, Sol. St. Electron. 13, 397 (1970).20S. Garbe, Sol. St. Electron. 12, 893 (1969).21S. Garbe, Phys. Stat. Sol. 2, 497 (1970).22J. J. Uebbing, L. W. James, J. Appl. Phys. 41, 4505

(1970).23L. W. James and J. J. Uebbing, Appl. Phys. Lett. 16,

370 (1970).24A. H. Sommer, H. H. Whitaker, and B. F. Williams,

Appl. Phys. Lett. 17, 273 (1970).25A. H. Sommer, J . Appl. Phys. 42, 2158 (1971).26S. Garbe, G. Frank, Gallium Arsenide and Related

Compounds, London, 1970; Proc. Intern. Conferenceon GaAs, Aachen, 1970.

27H. Schade, H. Nelson, and H. Kressel, Appl. Phys.Lett. 18, 121 (1971).

28A. F. Milton and A. D. Baer, J. Appl. Phys. 42, 5095(1971).

29J. M. Chen, Surf. Sci. 25, 457 (1971).30L. W. James, G. A. Antypas, J. Edgecumbe, R. L.

Moon, and R. L. Bell, J. Appl. Phys. 42, 4976 (1971).31E. M. Yee, and D. A. Jackson, Sol. St. Electron. 15,

245 (1972).32G. A. Allen, J. Phys. D4, 308 (1971).33RCA advertisement, Appl. Optics 10, No. 7, A19 (1971)

(photomultiplier with a GaAs cathode).MA. A. Mead, Sol. St. Electron 9, 1023 (1966).35J. J. Scheer and J. van Laar, Sol. St. Comm. 5, 303

(1967).36J. Van Laar and J. J. Scheer, Surf. Sci. 8, 342 (1967).37J. J. Uebbing, and R. L. Bell, Appl. Phys. Lett. 11, 357

(1967).38J. van Laar and J . J. Scheer, Philips Techn. Rev. 29,

54 (1968).39J. J. Scheer and J. van Laar, Surface Sci. 18, 130

(1969).40W. Monch, Surf. Sci. 21, 443 (1970).41L. W. Swanson and R. W. Strayer, J. Chem. Phys. 48,

2421 (1968); B. E. Evans, L. W. Swanson and A. E.Bell, Surf. Sci. 11, 1 (1968).

42J. M. Chen, J. Appl. Phys. 41, 5008 (1970).43T. E. Madey and J. T. Yates, J. Vac. Sci. Techn. 8, 39,

525 (1971).44W. Klein, Bull. Am. Phys. Soc. 14, 854 (1969).45J. J. Uebbing and N. J. Taylor, Bull. Am. Phys. Soc.

14, 792 (1969); N. J. Taylor, Vacuum 19, 575 (1969);J. J. Uebbing, J. Appl. Phys. 41, 802, 804 (1970);

J. Vac. Sci. Techn. 7, 81 (1970).46D. J. Miller and D. Haneman, Surf. Sci. 19, 45 (1970).47H. M. Love and H. D. Wiederick, J. Phys. E2, 1091

(1969).48D. L. Schaefer, Rev. Sci. Instr. 41, 274 (1970).49W. Klein, Rev. Sci. Instr. 42, 1082 (1971).50R. L. Bell and J. J. Uebbing, Appl. Phys. Lett. 12, 76

(1968); R. L. Bell, U.S. Patent No. 60144/68, Ser. No.1258917 (Cs2O on InP); British Patent (HOIj l/34) No.1258918 (InAsP).

51J. J. Uebbing and R. L. Bell, Proc. IEEE 56, 1624(1968).

52D. L. Schaefer, J. Appl. Phys. 40, 445 (1969).53W. Klein, J. Appl. Phys. 40, 4384 (1969); Bull. Am.

Phys. Soc. 14, 791, 855 (1969).54B. F. Williams, Appl. Phys. Lett. 14, 273 (1969).55R. E. Simon, A. H. Sommer, J. J. Tietjen, and B. F.

Williams, Appl. Phys. Lett. 15, 43 (1969); R. E. Simon,B. F. Williams, and R. Wasserman, British Patent(HOI j 1(32)) No. 1233721 (GaAsP).

56S. Garbe, Phys. St. Sol. 33, K87 (1969).57G. A. Antypas and L. W. James, J. Appl. Phys. 41,

2165 (1970).58H. Sonnenberg, Appl. Phys. Lett. 16, 245 (1970); 19,

431 (1971).59R. U. Martinelli, Appl. Phys. Lett. 16, 261 (1970).60L. W. James, G. A. Antypas, J. J. Uebbing, T. O.

Yep, R. L. Bell, J. Appl. Phys. 42, 580 (1971).61D. G. Fisher, R. E. Enstrom, and B. F. Williams,

Appl. Phys. Lett. 18, 371 (1971).62D. G. Fisher, R. E. Enstrom, J. S. Escher, and B. F.

Williams, J. Appl. Phys. 43, 3815 (1972).63R. L. Bell, L. W. James, G. A. Antypas, J. Edge-

cumbe, and R. L. Moon, Appl. Phys. Lett. 19, 513(1971).

64R. F. Steinberg, Appl. Phys. Lett. 12, 63 (1968).65S. Garbe and G. Frank, Sol. St. Comm. 7, 615 (1969).66B. J. Stocker, M. J. Plummer and A. A. Turnbull,

J. Phys.D3, 1299(1970).67T. Yasar, R. F. Steinberg, J. Vac. Sci. Techn. 6, 905

(1969); 7, 139 (1970).68Syms, CHA Adv. Electron. El. Phys. 28a, 406 (1969);

British Patent (HOI j 9/l2) No. 1193002, May 28,1970.

69G. A. Antypas, L. W. James, and J. J . Uebbing, J.Appl. Phys. 41, 2888 (1970).

70Y. Z. Liu, J. L. Moll, and W. E. Spicer, Appl. Phys.Lett. 17, 60 (1970).

71D. Andrew, J . P. Gowers, J. A. Henderson, M. J .Plummer, B. J. Stocker, and A. A. Turnbull, J. Phys.D3, 320 (1970).

72seeRef. 32.73S. B. Hyder, J . Vac. Sci. Techn. 8, 228 (1971).74J. C. Word, U.S. Patent cl. 313-95 (HOI j 39/l8) No.

3575628, Nov. 26, 1968 (GaAs on Si).75P. R. Sehvay, British Patent cl. HID (HOI j l/34) No.

1239893, Mar. 5, 1970 (GaAs on AlGaAs).76R. E. Simon and B. J . Williams, IEEE Trans. NS-15,

167 (1968).77R. E. Simon, A. H. Sommer, J. J. Tietjen, and B. F.. Williams, Appl. Phys. Lett. 13, 355 (1968).78R. E. Simon, B. F. Williams, and R. Wasserman, U.S.

Patent cl. 250-207 (HOI j) No. 3478213, Sep. 5, 1967,Nov. 11, 1969.

79G. A. Morton, H. M. Smith, and H. R. Krall, Appl.Phys. Lett. 13, 356 (1968); IEEE Trans. NS-16, 92(1969).

80W. A. Gutierrez, H. D. Pommerrenig and S. L. Holt,

737 Sov. Phys.-Usp., Vol. 16, No. 5, March-April 1974 N. A. Soboleva 737

Appl. Phys. Lett. 21, 249 (1972).81J. Llacer, IEEE Trans. NS-17, 29 (1970); Bull. Am.

Phys. Soc. 14, 854 (1969).82L. F. Afonina, A. I. Klimin, and G. B. Stuchinskii,

filektron. Tekhn. 4, 64 (1970).83L. F. Afonina, V. M. LagodinsMi, and G. B. Stuchin-

skii, Izv. AN SSSR, Ser. Fiz. 35, 1046 (1971).84R. U. Martinelli, Appl. Phys. Lett. 17, 313 (1970).85P. B. Coates, J. Phys. D7, L25 (1970).86Canad. El. Eng. 12, 31, 32 (1968); 13, 9 (1969) (photo-

multiplier with GaP dynodes); Phys. Today 23, (10)21 (1970).

87G. Present and D. B. Scral, Rev. Sci. Instr. 41, 771(1970).

88H. R. Krall, F. A. Helvy, and D. E. Persyk, IEEETrans. NS-17, 71 (1970).

89B. F. Williams and R. E. Simon, Appl. Phys. Lett. 14,214 (1969).

90E. S. Kohn, Appl. Phys. Lett. 18, 272 (1971)."Electronics 44, 29 (1971); Sol. St. Technol. 14, 35

(1971).92H. Kressel, E. S. Kohn, H. Nelson, J. J. Tietjen, and

L. R. Weisberg, Appl. Phys. Lett. 16, 359 (1970); Elec-tronics 43, 45 (1970).

93H. Schade, H. Nelson, and H. Kressel, Appl. Phys. Lett.

18,413(1971).94V. L. Dalai, J. Appl. Phys. 43, 1160 (1972).95S. A. Ward, U.S. Patent cl. HOI (j, 7/54) No. 3591424,

Jun. 26, 1969, Jul. 6, 1971.96B. V. Dore and D. V. Geppert, U.S. Patent cl. 313-94

No. 3408521, Nov. 22, 1965, Oct. 29, 1968.97A. G. Milnes and D. L. Feucht, Appl. Phys. Lett. 19,

383 (1971).98E. D. Savoy and J. J. Tietjen, Laser Focus 6, 34 (1970).99H. Sonnenberg, IEEE J. Sol. St. Circ. 5, 272 (1970).

100N. A. Soboleva, Effektivnye fotoemittery (EfficientPhotoemitters), VINITI (deposited document), 1970.

101R. L. Bell and W. E. Spicer, Proc. IEEE 58, 1788(1970).

102N. N. Petrov, Zh. Tekh. Fiz. 41, 2473 (1971) [Sov.Phys.-Tech. Phys. 16, 1965 (1972)].

103L. Eckertova, V. Stary, Vyzkumne zpravo, 1971.104B. F. Williams, J. J. Tietjen, Proc. IEEE 59, 1489

(1971).105M. Jedliffka, Slaloproudy obzor 33, 44 (1972).106B. F. Williams, IEEE Trans. NS-19, 39 (1972).

Translated by M. V. King

738 Sov. Phys.-Usp., Vol. 16, No. 5, March-April 1974 N. A. Soboleva 738