"Study of Indium Tin Oxide (ITO) for Novel Optoelectronic Devices"
Ph.D. thesis by Shabbir A Bashar




2.4 Heterojunction Bipolar Transistors

The idea behind the heterojunction bipolar transistor (HBT) is as old as the transistor itself. It was proposed by W. Shockley in 1948 and subsequently patented by him in 1951 [36]. However it was H. Kromer [37], who is credited with developing the detailed theory and analysis of HBTs leading up to its present day status in semiconductor electronics and integrated circuit applications.


2.4.1 Heterostructure Concepts

Kromer's Central Design Principle (CDH) of heterostructure devices uses the energy gap variations in addition to electric fields as forces acting on electrons and holes, to control their distribution and flow [38]. Therefore, by carefully selecting the appropriate bandgap and the electric fields, it becomes possible (within limits) to control the forces acting on electrons and holes, separately and independently of each other - a design freedom not rendered by homostructures. With the evolution of more and more advanced growth technologies such as MBE and MOCVD, a change in semiconductor (and hence in energy gap) is not significantly harder to achieve than a change in doping level.

In a practical HBT, the emitter consists of a wide bandgap material (e.g. AlGaAs, InGaP or InP) while the base is a narrow bandgap material (e.g. GaAs or InGaAs). The band energy diagram for such an emitter-base heterojunction, before and after the two materials are brought into contact, is depicted in Figure 2.4a and b respectively.

Figure 2.4: Energy band diagram of a wide bandgap N-type emitter and a narrow bandgap p- type base heterojunction at equilibrium (a) before and (b) after formation.

As in homojunctions, once contact is made the conduction and valence bands join in a manner such that the Fermi levels align. However, since the electron affinities of the emitter and the base, cE and cB respectively, are different in heterojunctions a discontinuity in the conduction band, DEc, appears. Analytically, this is given by:

DEc = cB - cE (eqn. 2.17)

Since, the bandgap energies are different, a discontinuity in the valence band, DEv, also appears. The total bandgap discontinuity, DEg, is the sum of the discontinuities in the conduction and valence bands:

DEg = DEc + DEv (eqn. 2.18)

In the HBT a large DEc is undesirable as it acts as an extra barrier to electron injection from the emitter to the base thereby requiring a high emitter-base forward bias voltage, Vbe, (typically 1.3V) [39]. This ultimately limits the device current gain far below that expected from an HBT [38]. By the same analogy, a large DEv is ideal as it limits reverse injection of holes from the base into the emitter. Table 2.2 lists some of the band discontinuities of common HBT systems. Thus a large DEv/DEc ratio is desirable.

HBT Material Structure DEc (eV) DEv (eV) DEg (eV) DEv/DEc Ref. No.
Al0.3Ga0.7As/GaAs 0.24 0.13 0.37 0.54 [40]
In0.5Ga0.5P/GaAs 0.19 0.29 0.48 1.53 [41,42,43]
InP/In0.53Ga0.47As 0.25 0.34 0.59 1.36 [44]

Table 2.2: Band discontinuities for common HBT material heterostructures

Heterojunctions, such as InGaP/GaAs which naturally exhibit relatively large DEv compared to their DEc, are more advantageous than AlGaAs/GaAs. However, since AlGaAs is lattice matched to GaAs for all fractions of Al, it is possible to reduce or even remove this conduction band discontinuity by compositionally grading the mole fraction of Al in the emitter material immediately (few hundred angstroms) before the GaAs base [45]. This helps lower the Vbe by 150mV compared to that in the non-composition graded emitter HBTs. However, it should be noted that the bandgap of AlGaAs becomes indirect once the Al mole fraction is raised above 50%, and thus is not of interest from a device application point of view.

Using Figure 2.4 the built-in potential, Vbi, for a heterostructure can be derived as follows:

EgE - Vn = DEv + Vp + Vbi (eqn. 2.19)

where,
EgE = emitter bandgap in eV
Vn = potential difference between the bottom of conduction band and Fermi level
Vp = potential difference between Fermi level and the top of valence band

Expressing Vn and Vp in terms of emitter and base doping levels, NDE and NAB, and the respective density of states in emitter conduction band and base valence band, NcE and NvB, we obtain:

Vn = (EcE - EfE) = - (kT/q) ln (NDE/NcE) (eqn. 2.20)

and,

Vp = (EfB - EvB) = - (kT/q) ln(NAB/NvB) (eqn. 2.21)

Combining (eqn. 2.18) to (eqn. 2.21), we obtain:

Vbi = EgE + (kT/q) ln(NDE/NcE) - DEg + DEc + (kT/q) ln(NAB/NvB) (eqn. 2.22)

But,

DEg = EgE - EgB (eqn. 2.23)

Hence,

Vbi = EgB + DEc + (kT/q) ln(NDE.NAB/NcE.NvB) (eqn. 2.24)


2.4.2 Current Transport in HBTs

The operating principle of a (n-p-n) bipolar transistor consists of electron injection from the emitter into the base and their subsequent collection by the collector. Figure 2.5 shows the energy band structure of an N-p-n HBT with wide-gap emitter, with the various current components, and the hole repelling effect of the additional energy gap in the emitter. The band energies are expressed in joules; thus bandgap is expressed using Wg rather than Eg. In this case it is assumed that the emitter junction has been graded sufficiently to obliterate any band edge discontinuities or even any non-monotonic variations of the conduction band edge.

Figure 2.5: Band diagram of an N-p-n HBT biased in emitter ground configuration

There are the following injection related dc currents flowing in such a transistor:

  1. A current In of electrons injected from the emitter into the base;
  2. A current Ip of holes injected from the base into the emitter;
  3. A current Iscr due to electron-hole recombination within the forward biased emitter-base space-charge layer.
  4. A small part, Ibulk, of the electron injection current, In, is lost due to bulk recombination in the base.
  5. A small thermally generated minority hole current from the n-type collector, Icbo, flows into the base from the reverse biased base-collector junction.

Emitter current, Ie = In + Ip + Iscr (eqn. 2.25)
Collector Current, Ic = In - Ibulk + Icbo (eqn. 2.26)
Base Current, Ib = Ip + Ibulk + Iscr - Icbo (eqn. 2.27)

The device operation mainly depends on current contribution In while the other components are strictly nuisance currents, as are capacitive currents (not shown above) that accompany voltage changes. The effect of the space charge recombination current, Iscr, on the transistor performance has been explained elsewhere by Morgan et al [46]. Since the base thickness, wb, is much smaller than the electron diffusion length, LnB, in the base, Ibulk is very small. Similarly, any currents due to electron-hole pair generation in the collector depletion layer, Icbo, are also insignificant and can be ignored at this stage. Expressed in terms of the dc current gain, b:

b = Ic/Ib = (In - Ibulk + Icbo)/(Ip + Ibulk + Iscr - Icbo) (eqn. 2.28)

and,

bmax = In/(Ip + Iscr) (eqn. 2.29)

bmax is the highest possible value of b in the limit of various negligible recombination and other minor currents; in fact, it is the improvement in bmax to which the wide-gap emitter idea addresses itself. This can be expressed in terms of the electron and hole injection current densities, Jn and Jp [38]:

Jn = (NDE.vnB) exp(-qfn/kT) (eqn. 2.30)

Jp = (NAB.vpE) exp(-qfp/kT) (eqn. 2.31)

where,
qfn and qfp = potential energy barrier heights for electrons and holes at the B-E junction
vnB and vpE = mean speeds (due to the combined effects of drift and diffusion) of electrons at the emitter end of the base and holes at the base end of the emitter respectively.

The above equations are often expressed in terms of diffusion current effects only:

Jn = (qNDE.DnB/wb).exp(-qfn/kT) (eqn. 2.32)

Jp = (qNAB.DpE/LpE).exp(-qfp/kT) (eqn. 2.33)

where,
DnB, DpE = minority carrier diffusion coefficients in the base and emitter respectively
LpE = hole diffusion length in the emitter region
If the band gap of the emitter is larger than that of the base by DWg, we have :

q(fp - fn) = DWg (eqn. 2.34)

Hence,

In/Ip = bmax = (NDE.vnB/NAB.vpE) exp(DWg/kT) (eqn. 2.35)

or in terms of diffusion currents only,

bmax = (NDE.DnB.LpE/NAB.DpE.wb) exp(DWg/kT) (eqn. 2.36)

Of the three variables in (eqn. 2.35), vnB/vpE is least subject to manipulation; hence, for a large bmax, we need either NDE >> NAB or DWg to be several times larger than kT. Since bandgap differences many times greater than kT is readily obtainable, very high values of In/Ip can be achieved almost regardless of the doping ratio. Therefore, the hole injection current Ip becomes negligible and Ib » Iscr + Ibulk. For a useful, transistor we must still have Ibulk << In. If we approximate Ie by In, we obtain:

b = In/(Iscr + Ibulk) (eqn. 2.37)

In contrast to homojunction transistors, in a properly designed high gain HBT, the interface recombination current component Iscr, can be as small as In/1e3 making Ibulk the dominant part of Ib. Hence, Iscr is neglected beside Ibulk; the bulk recombination current density may be written as [38]:

Jbulk = gne(0) (wb/tn) (eqn. 2.38)

where,
ne(0) = the injected electron concentration at the emitter end of the base
tn = average electron life time in the base
g = a factor between 0.5 and 1 indicating how much electron concentration in the base
differs from the electron concentration at the emitter end of the base.

Thus combining (eqn. 2.30), (eqn. 2.37) and (eqn. 2.38) and making the necessary substitutions, we may write the gain as follows:

b = (1/g)(vnB.tn/wb) (eqn. 2.39)

This depends on the base doping only through the effect of the base doping on the lifetime; for heavy base doping levels, the life times may be short [47]. Evidently no serious problems from reduced minority carrier lifetimes arise unless the latter drop to the vicinity of 1e-10 s or lower, at least not for plausible base widths not exceeding 1000Å.


2.4.3 Fundamental Advantages of HBTs

Some of the key benefits of HBTs can be summarised as follows [48]:


2.4.4 Abrupt Emitter-Base HBT

A conduction band discontinuity, DEc, exists in the abrupt emitter-base heterojunction as shown in Figure 2.6. Since in a practical HBT, NAB >> NDE, this discontinuity appears close to the base leading to a potential barrier DEb. In addition there is also an insignificant conduction band notch on the base side, DEn; thus, DEc » DEb.

Figure 2.6: Band diagram of an abrupt emitter/base heterojunction

Hence, for an abrupt junction, (eqn. 2.34) is modified to:

q(fp - fn) = q(DEc + DEv - DEb) » q(DEv) = DWv (eqn. 2.40)

Hence (eqn. 2.35) is also modified and the b for a HBT with an abrupt emitter-base junction is now given by:

babrupt = (NDE.vnB/NAB.vpE) exp(DWv/kT) (eqn. 2.41)

or,

babrupt = (NDE.DnB.LpE/NAB.DpE.wb) exp(DWv/kT) (eqn. 2.42)


2.4.5 Common III-V HBT Material Systems

Among all the GaAs-based materials used for HBTs, the most investigated combination has been the AlGaAs/GaAs system which exhibits high current gain, high fT and fmax. However, the InP-based lattice matched systems, in particular InP/In0.53Ga0.47As came into contention primarily because of its bandgap of 0.75eV which is sensitive to 1300nm and 1550nm wavelengths - wavelengths which correspond to the low loss windows of optical fibers and universally used for long haul optical fiber telecommunications. Apart from the high current driving capacity of HBTs, the material compatibility that the InP/In0.53Ga0.47As system allows, one can monolithically integrate with HBTs with other optoelectronic devices. Also its superior transport properties with respect to the AlGaAs/GaAs system makes it one of the strongest competitors.

The material band alignments for various materials for HBTs were shown in Table 2.2. DEv for the InP/InGaAs heterojunction is much larger than its DEc allowing a high injection efficiency without compromising the base doping level. Extremely high doping (NAB >1e20 cm-3) has been achieved in InGaAs by GSMBE using Be as the p-type base dopant [49]. High frequency performance is also expected for InGaAs/InP HBTs due to the high electron mobility of InGaAs. In addition, the large G-X and the G-L inter-valley separation (DEG-X =1.0eV, DEG-L =0.55eV ) in InGaAs allows large collector voltage swing with high peak carrier velocity. Furthermore, the surface recombination velocity associated with InGaAs, which is 1e3 lower compared to that of GaAs [50], enables the development of transistors with high current gain but insensitive to collector current density and emitter dimension; thus allowing the latter to be scaled to the sub-micrometer regime.


© 1998: Shabbir A. Bashar (in accordance with paragraph 8.2d, University of London Regulations for the Degrees of M.Phil. and Ph.D., October 1997). The Copyright of this thesis rests with the author, and no quotation from it or information derived from it may be published without the prior written consent of the author.
Title Dedication Acknowledgements Abstract Symbols Contents

This page is hosted by Dr. Shabbir A. Bashar, e-mail: shabbir@betelco.com