SEMICONDUCTOR HETEROJUNCTIONS A p-n junction is formed when a p-type doped portion of the semiconductor is d with an n-type doped portion. As a fundamental component for functions such as rectification, the p-n junction forms the basic unit of a bipolar transistor. If both the p-type and the n-type regions are of the same semiconductor material, the junction is called a homojunction. If the junction layers are made of different semiconductor materials, it is a heterojunction. As a matter of convention, if the n-type doped semiconductor material has larger energy gap than the p-type doped material, it is denoted a p-N heterojunction. The use of capital and lowercase Ietters connotes the relative size the energy gap. Conversely, if the p-type doped material has a larger energy gap than the n-type material, the junction is referred to as a P-n heterojunction. © Nezih Pala npala@fiu.edu Semiconductor Device Theory EEE 6397 – 19 Heterojunctions -2 When semiconductors of different band gaps, work functions, and electron affinities are brought together to form a junction, we expect discontinuities in the energy bands as the Fermi levels line up at equilibrium . The discontinuities in the conduction band EC and the valence band EV accommodate the difference in band gap between the two semiconductors Eg. In an ideal case, EC would be the difference in electron affinities q(2 - 1), and Ev would be found from Eg - EC This is known as the Anderson affinity rule. Eg Eg1 Eg 2 EC EV In practice, the band discontinuities are found experimentally for particular semiconductor pairs. © Nezih Pala npala@fiu.edu Semiconductor Device Theory EEE 6397 – 20 Heterojunctions -4 To draw the band diagram, we need: depend on semiconduc tor electron affinity ( ) material, NOT doping band gap (E g ) and depends on semiconduc tor Work function material AND doping The electron affinity and work function are referenced to the vacuum level. The electron affinity and work function are referenced to the vacuum level. The true vacuum level (or global vacuum level), Evac, is the potential energy reference when an electron is taken out of the semiconductor to infinity, where it sees no forces. Hence, the true vacuum level is a constant © Nezih Pala npala@fiu.edu Semiconductor Device Theory EEE 6397 – 21 Heterojunctions -5 However, since the electron affinity is a material parameter and therefore constant we need to introduce the new concept of the local vacuum level, Evac(loc), which varies along with and parallel to the conduction band edge, thereby keeping the electron affinity constant. The local vacuum level tracks the potential energy of an electron if it is moved just outside of the semiconductor, but not far away. P-side n-side The difference between the local and global vacuum levels is due to the electrical work done against the fringing electric fields of the depletion region, and is equal to the potential energy qV0 due to the built-in contact potential V0 in equilibrium. This potential energy can, of course, be modified by an applied bias. © Nezih Pala npala@fiu.edu Semiconductor Device Theory EEE 6397 – 22 Heterojunctions -6 To draw the band diagram for a heterojunction accurately, we must not only use the proper values for the band discontinuities but also account for the band bending in the junction. To do this, we must solve Poisson's equation across the heterojunction, taking into account the details of doping and space charge, which generally requires a computer solution. We can, however, sketch an approximate diagram without a detailed calculation. Given the experimental band offsets EV and EC, we can proceed as follows: 1. Align the Fermi level with the two semiconductor bands separated. Leave space for the transition region. © Nezih Pala npala@fiu.edu Semiconductor Device Theory EEE 6397 – 23 Heterojunctions -7 2. The metallurgical junction (x = 0) is located near the more heavily doped side. At x = 0 put EV and EC, separated by the appropriate band gaps. 3. Connect the conduction band and valence band regions, keeping t he band gap constant in each material. Steps 2 and 3 of this procedure are where the exact band bending is important and must be obtained by solving Poisson's equation. In step 2 we must use the band offset values EV and EC for the specific pair of semiconductors in the heterojunction. © Nezih Pala npala@fiu.edu Semiconductor Device Theory EEE 6397 – 24 Heterojunctions -8 Consider heavily n-type AlGaAs is grown on lightly doped GaAs where the discontinuity in the conduction band allows electrons to spill over from the N+AlGaAs into the GaAs, where they become trapped in the potential well. As a result, electrons collect on the GaAs side of the heterojunction and move the Fermi level above the conduction band in the GaAs near the interface. These electrons are confined in a narrow potential well in the- GaAs conduction band. If we construct a device in which conduction occurs parallel to the interface, the electrons in such a potential well form a two-dimensional electron gas with very negligible impurity scattering in the GaAs well, and very high mobility controlled almost entirely by lattice scattering (phonons). © Nezih Pala npala@fiu.edu Semiconductor Device Theory EEE 6397 – 25 p+-N Heterojunction under equilibrium -1 Consider, for example, a p-N heterojunction formed with a p-type GaAs layer doped at 3x1019 cm-3 and an N-type AIGaAs layer doped at 1x1016 cm-3 The band diagrams of individual layers are shown in the figure side by side with their vacuum levels aligned to illustrate the difference in the Fermi levels in the chargeneutral condition . © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 26 p+-N Heterojunction under equilibrium -2 When the two regions are brought into contact, the Fermi level on the N side is initially higher than that on the p-side. Therefore, the electrons in a p-N junction lend to flow from regions of higher Fermi level (N regions) to regions or lower Fermi levels (p regions). On the other hand, the holes flow from the region with a lower Fermi level toward the region with a higher Fermi level. Therefore, the holes flow from the p- type layer toward the N-type layer consistent with the fact that there are more holes in the p-type region and they tend to diffuse to the region with less holes. As these mobile carriers move toward the other sides, they leave behind the uncompensated dopant atoms near the junction. In the p-type region, the uncompensated acceptors are negative ions. In the N-type region, the uncompensated donors are positive ions. Therefore, the carrier diffusion results in a electric field pointing from the N-type region to the p-type region. This electric field retards further electron diffusion from the N-type toward the p-type as well as hole diffusion in the opposite direction. Hence, there is a natural negative feedback mechanism such that the more carrier movement takes place, the larger the electric field becomes and the tendency of the carrier movement is reduced. © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 27 p+-N Heterojunction under equilibrium -3 Eventually, an equilibrium condition sets in and the tendency of the carrier diffusion is exactly counterbalanced by the electric field that impedes the carrier movement. At this precise movement, the Fermi levels at the two regions are aligned. The above exchange of carriers occurs on a very short time scale. So the whole process can be thought of as instantaneous. The two side immediately adjacent to the junction where the dopants become uncompensated are called the depletion region, meaning that they are depleted of mobile carriers. (Another name for the depletion region is the space-charge region). At the two extreme ends away from the actual junction, carrier movement has never occurred. Their dopant atoms are still compensated by their respective electrons or holes. These are called the neutral regions because the net charge concentrations there are zero. © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 28 p+-N Heterojunction under equilibrium -3 Fermi level during thermal equilibrium lines up throughout the entire semiconductor. At regions far away from the junction where the semiconductor remains neutral, the relative positions of Ef with respect to Ec and to Ev are unmodified from those prior to the joining of the two sides. The conduction and valence band edges at the sides are not at the same level, since it is the Fermi level that must be aligned. The conduction and valence band edges across the depletion region must be connected somehow to form continuous curves. © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 29 p+-N Heterojunction under equilibrium -4 The Poisson equation must be solved to ascertain the manner in which the conduction and valence band edges are connected across the depletion region: d q ( p n Nd Na ) dx S Let us define that x = 0 correspond to the junction boundary, with the p side in the -x direction and the N side in the + x direction. Moreover, x = - Xpo is the boundary separating the neutral and the depletion regions on the p side, and XN0 the boundary on the N side. The subscript 0 emphasizes that we are considering the thermal equilibrium condition. Considering that, in the p-side depletion region, the net charge concentration is the negative acceptor density and in the N-side depletion region the net charge concentration is the donor density (depletion approximation). Therefore, the last equation simplifies to d q N a for -X p 0 x 0 dx p d q Nd dx N © Nezih Pala npala@fiu.edu for 0 x X N 0 EEE 6397 – Semiconductor Device Theory 30 p+-N Heterojunction under equilibrium -5 After integrating and applying the boundary conditions that the electric fields at – Xp0 and at XN0 are zero, the electric field within the depletion region can be found as ( x) q N a (x X p 0 ) for -X p 0 x 0 p ( x) q N d ( X N 0 - x) N for 0 x X N 0 In the p-side, the potential profile is obtained by integrating (x) q x2 V ( x) ( x)dx N a ( X p 0 x) C for -X p 0 x 0 p 2 Since it is the relative value or potentials rather than their absolute values that is of importance, we arbitrarily define the zero potential at a convenient location, namely, at x =-Xpo .With the boundary condition that V(- Xpo) = 0, V(x) is written as 2 X p0 q x2 V ( x) Na ( X p0 x ) p 2 2 © Nezih Pala npala@fiu.edu for -X p 0 x 0 EEE 6397 – Semiconductor Device Theory 31 p+-N Heterojunction under equilibrium -6 The built-in potential on the p side (p0 ) is the difference in the potentials at x=0 and x=-Xp0. It is readily verified that p0 q N a X 2p0 2 p where p0 is a positive number. The band diagram, being an energy diagram for the negatively charged electron, shows that the electron energy decreases with the associated increase in V(x). Similarly the potential profile on the N side can be obtained by integrating the appropriate electric profile. Taking the boundary condition that V(0) = p0, q x2 V ( x) N d X N 0 x - p0 N 2 for 0 x X N 0 The built-in potential on the N side ( N0 ) is the difference in the potentials at x = 0 and x = XN0. V(XN0) - V(0) is then equal to N 0 © Nezih Pala npala@fiu.edu q N d X 2N0 2 N EEE 6397 – Semiconductor Device Theory 32 p+-N Heterojunction under equilibrium -7 The overall built-in junction potential (bi ) is the potential difference from the neutral region on one side the neutral region on the other. It is equal to bi N 0 p 0 q q N d X 2N0 N a X 2p0 2 N 2 p Xp0 and XN0 can be determined by solving two linearly independent equations relating these two variables. One of the required equations is obtained by enforcing the continuity of the electric flux density D = in the absence of an interfacial charge density at the junction: N d X N0 N a X p0 The equality ensures that charge neutrality exists in the overall p-N junction. The second equation relating Xp0 and XN0 is found by taking the ratio of the built ·in potentials on the N side to those on the p side: N 0 p N d X N0 p 0 N N a X 2p0 2 © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 33 p+-N Heterojunction under equilibrium -8 Substituting the charge conservation relationship into the above equation, we rewrite the ratio as N 0 p N a p 0 N N d From this equation, the built-in potential of the junction is N 0 p N a p 0 bi 1 p 0 1 N N d p0 Alternatively, from an examination of the band diagram bi is equal to bi Egp Ec p N q where Egp is the energy gap of the p-side material © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 34 p+-N Heterojunction under equilibrium -9 p and N represent the differences of the Fermi levels with respect to the valence band edge and the conduction band edge, respectively, in the neutral regions. That is, p E f Ev | x N Ec E f | x Both p and N are calculated from the equilibrium statistics equations given in n Nce E f Ec kT p Nve Ev E f kT Once the doping levels are specified, bi for the heterojunction is readily calculated. Other quantities such as N0 and V(x) are then obtained from the Equations derived above. © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 35 p+-N Heterojunction under equilibrium -10 EXAMPLE Let us draw a band diagram for a (p) GaAs/(N) AI0.35Ga0.65As. As heterojunction at thermal equilibrium. The doping level and relevant material parameters are Egap (GaAs) =1.42 eV Egap (AlAs) =2.16 eV Nd=1x1016 cm-3 Na=3x1019 cm-3 n NC e p E f Ec kT Ev E f kT n n E f Ec kT ln Nd p ln NV p A1 NV p A2 NV 3.72 1017 0.0259 ln 0.093eV 16 110 2 p A3 NV 3 p A4 NV 4 0.103eV With A1=3.53x10-1, A2=-4.95x10-3, A3=1.48x10-4, A4=-4.42x10-6 Nv=4.7x1018 cm-3 © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 36 p+-N Heterojunction under equilibrium -11 Energy gap of the AlxGa1-xAs material is determined by the valley at x<0.45, but is determined by the X valley at x>0.45. The energy gap (in eV) is given as: 1.424 1.247 x for 0 x 0.45 Eg ( Al xGa1 x As ) 1.86eV 2 1.90 0.125 x 0.143x for x 0.45 The energy gap difference (Eg) in an AlxGa1-xAs/GaAs heterojunction is shared between the conduction band and the valence band. That is Eg= Ec+ Ev where Ec is the conduction band discontinuity and Ev is the valence band discontinuity. The amount of Ev is linearly dependent on the aluminum mole fraction (x) for all values of x. Ev and Ec in an AlxGa1xAs/GaAs heterojunction are therefore given by Ev ( x) 0.55x 0 x 0.45 1.247 0.55 x Ec ( x) 2 0 . 476 ( 0 . 125 0 . 55 ) x 0 . 143 x x 0.45 E0. is 0.244 eV. Egp the energy gap of the narrower gap material (GaAs), is 1.424 eV. Therefore, the built-in voltage is found bi 1.424 0.244 (0.103) 0.093 1.678eV © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 37 p+-N Heterojunction under equilibrium -12 The relative dielectric constants are 13.18 for GaAs and 12.09 for Al0.35Ga0.65As. p0 can be found as p N a p N a bi 1 p 0 p 0 bi /1 N N d N N d 13.18(3 1019 ) 4 1.678 / 1 5 . 13 10 V 16 12.09(110 ) N0 which is equal to bi - p0 is 1.678 - 5. 13 x 10-4 1.6775 V. It is clear from this calculation that if one side is significantly more heavily doped than the other practically all of the built in potential drops across the depletion region of the lightly doped side. Once N0 and p0 are determined, XN0 and Xp0 are calculated from N 0 2 N q 2(12.09)(8.85 1014 ) 2 -5 N d X N0 X N0 N 0 1 . 677 4.74 10 cm 19 16 2 N qN d (1.6 10 )(110 ) p0 2 p q 2(13.18)(8.85 1014 ) 2 N a X p0 X p0 p0 5.13 104 1.58 10-8 cm 19 19 2 p qN a (1.6 10 )(3 10 ) © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 38 p+-N Heterojunction under equilibrium -13 The depiction thickness of the heavily doped side is merely 1.58 which is practically zero. With the calculated parameters, the band diagram is drawn as shown in the figure. The band profiles vary parabolically with position as given by equations: 2 X p0 q x2 V ( x) Na ( X p0 x ) p 2 2 q x2 V ( x) N d X N 0 x - p0 N 2 © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 39 Semiconductor Heterojunctions Abrupt p-n heterojunction N-type p-type To construct a band diagram for an abrupt p-n heterojunction, we proceed in the same manner as for the p-n homojunction. We begin with two separate materials. Since the materials have different bandgaps, there must exist a discontinuity in the conduction band (ΔEc) and/or the valence band (ΔEv) at the interface . The difference in the bandgap between the two materials is equal to the sum of the conduction band and valence band discontinuities Eg Eg1 Eg 2 Ec Ev © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 40 Graded p-n heterojunction Although the abrupt p-n heterostructure discussed in the previous section did result in an increase in the barrier to hole injection, the notch in the conduction band at the interface also caused an undesirable increase in the barrier to electron injection. While the net effect was still an increase in the ratio In/Ip, eliminating this notch further increases In/Ip to a value In DN L E n d p exp g k T I D N L p a n B p het © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 41 Graded p-n heterojunction -2 In order to reduce this notch, the bandgap of the p material can be graded upwards from the junction, as shown in the figure. For example, in an n-AlGaAs/p-GaAs graded heterostructure, the n material is GaAs at the junction and is graded to the final AlGaAs composition over a short distance. The final shape of the notch depends on the length and profile of the grade; longer grading typically gives a smaller notch. However, it is important that the grade is contained well within the depletion region. If the grade ends outside the depletion region, then the barrier seen by holes decreases, thus reducing the benefits of the heterojunction. Note that the barrier to holes in both abrupt and graded heterojunctions is the same. It is just the barrier for electron flow that is reduced in the graded structures, allowing for the increased ratio of In/Ip. © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 42 Example: Designing a p-n heterojunction grade Consider four different n-p+ Al0.3Ga0.7As/GaAs heterojunctions with ND = 1017 and NA = 5× 1018. The AlGaAs in these junctions is graded from x = 0 to x = 0.3 over XGrade = 0(abrupt), XGrade = 100°A, XGrade = 300°A, and XGrade = 1μ. Calculate and plot the energy band diagrams for the above four cases. © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 43 Example: Designing a p-n heterojunction grade © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 44 Quasi-electric fields In a homogeneous semiconductor, the separation between the conduction and valence bands is everywhere equal to the semiconductor bandgap. Any electric field applied to the material therefore results in an equal slope in the conduction and valence bands, as indicated in the figure. When a hole or electron is placed in this structure, a force of magnitude eE will act on the particle. The magnitude of the force is equal to the slope of the bands and is the same for both electrons and holes. However, the direction of the force is opposite for the two particles. © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 45 Quasi-electric fields -2 An interesting phenomenon arises in semiconductors with graded bandgaps, such as the bipolar transistor emitter-base structure shown at the top figure . In the graded region, the bandgap is not constant, so the slopes in the conduction and valence bands are no longer equal. Hence the forces acting on electrons and holes in this region are no longer equal in magnitude. It is in general possible for a force to act on only one type of carrier, as shown in the top figure, or for forces to act in the same direction for both electrons and holes, as in the bottom figure. Such behavior cannot be achieved by pure electric fields in homogeneous materials. These fields, which were first described by Herbert Kroemer in 1957, are therefore referred to as quasi-electric fields. © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 46 Quasi-electric fields -3 In a given material, the total field acting on a hole or an electron is always the sum of the applied field and the quasi field, or e,tot app e,quasi h,tot app h,quasi The applied field, which results from applying a voltage difference between the ends of the material, will always be the same for both electrons and holes, but the quasi field could be different for both. The band profiles in figures in the last slide can therefore be achieved in a number of different ways. For example, the profile in top figure could be achieved in the following two ways: 1. An undoped (intrinsic) material with a graded composition and zero applied electric field typically results in the profile in the lower figure. If an electric field app = −e,quasi is then applied to this material, the resulting profile will be the one shown in the upper figure. 2. A uniformly doped n-type material with a graded composition and zero applied electric field will also result in the profile in the upper figure. In this case, the doping ensures that the separation between the conduction band and the Fermi level remains approximately constant. Notice that the resulting quasi-electric field in this structure acts only on minority carriers. © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 47 Quasi-electric fields -4 Quasi-electric fields provide engineers with additional tools that can be exploited in device design. They have proven to be very useful in decreasing transit times in devices that rely on minority carrier transport. For example, in bipolar technology, a highly doped graded base layer is often used to speed up the transport of minority carriers from the emitter to the collector. For a base with uniform bandgap, minority carriers injected from the emitter must diffuse across the base, a process that is generally slow. By using a highly doped graded base to generate a quasi-electric field, such as was described in the second example above, minority carriers can be swept across much more quickly, thus reducing the base transit time and improving the device RF performance. © Nezih Pala npala@fiu.edu EEE 6397 – Semiconductor Device Theory 48