ArticlePDF Available

# Antimonide-based compound semiconductors for electronic devices: A review

Authors:
• Washington-Liberty High School

## Abstract and Figures

Several research groups have been actively pursuing antimonide-based electronic devices in recent years. The advantage of narrow-bandgap Sb-based devices over conventional GaAs- or InP-based devices is the attainment of high-frequency operation with much lower power consumption. This paper will review the progress on three antimonide-based electronic devices: high electron mobility transistors (HEMTs), resonant tunneling diodes (RTDs), and heterojunction bipolar transistors (HBTs). Progress on the HEMT includes the demonstration of Ka- and W-band low-noise amplifier circuits that operate at less than one-third the power of similar InP-based circuits. The RTDs exhibit excellent figures of merit but, like their InP- and GaAs-based counterparts, are waiting for a viable commercial application. Several approaches are being investigated for HBTs, with circuits reported using InAs and InGaAs bases.
Content may be subject to copyright.
Review
Antimonide-based compound semiconductors
for electronic devices: A review
Brian R. Bennett
*
, Richard Magno, J. Brad Boos, Walter Kruppa, Mario G. Ancona
Electronics Science and Technology Division, Naval Research Laboratory, Washington, DC 20375-5347, USA
Received 14 April 2005; accepted 11 September 2005
Available online 4 November 2005
The review of this paper was arranged by Prof. C. Tu
Abstract
Several research groups have been actively pursuing antimonide-based electronic devices in recent years. The advantage of narrow-
bandgap Sb-based devices over conventional GaAs- or InP-based devices is the attainment of high-frequency operation with much lower
power consumption. This paper will review the progress on three antimonide-based electronic devices: high electron mobility transistors
(HEMTs), resonant tunneling diodes (RTDs), and heterojunction bipolar transistors (HBTs). Progress on the HEMT includes the dem-
onstration of Ka- and W-band low-noise ampliﬁer circuits that operate at less than one-third the power of similar InP-based circuits. The
RTDs exhibit excellent ﬁgures of merit but, like their InP- and GaAs-based counterparts, are waiting for a viable commercial application.
Several approaches are being investigated for HBTs, with circuits reported using InAs and InGaAs bases.
Published by Elsevier Ltd.
PACS: 81.05.Ea; 81.15.Hi; 85.30.Pq; 85.30.Tv; 73.61.Ey; 73.63.Hs
Keywords: Antimonide; HEMT; RTD; HBT; MBE; InAs
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1876
2. InAs quantum wells and high-electron-mobility transistors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1876
2.1. Background . ......................................................................... 1876
2.2. InAs/AlSb single quantum wells ............................................................ 1877
2.3. Doping of HEMTs ..................................................................... 1877
2.4. HEMT design and performance ............................................................ 1878
2.5. Variations to standard HEMT design ........................................................ 1881
2.6. Microwave and low-frequency noise . ........................................................ 1881
2.7. Radiation effects ....................................................................... 1882
2.8. Circuits ............................................................................. 1882
2.9. InSb-channel HEMTs . . ................................................................. 1883
3. Resonant tunneling diodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1883
3.1. Background . ......................................................................... 1883
3.2. Performance: peak current density and peak-to-valley ratio . ....................................... 1884
0038-1101/- see front matter Published by Elsevier Ltd. doi:10.1016/j.sse.2005.09.008 * Corresponding author. Tel.: +1 202 767 3665; fax: +1 202 767 1165. E-mail address: brian.bennett@nrl.navy.mil (B.R. Bennett). www.elsevier.com/locate/sse Solid-State Electronics 49 (2005) 1875–1895 3.3. Bias voltages and hysteresis ............................................................... 1886 3.4. Substrates, defects, and interfaces ........................................................... 1887 3.5. Radiation effects ....................................................................... 1887 3.6. Applications . ......................................................................... 1888 4. Heterojunction bipolar transistors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1889 4.1. Background and early work ............................................................... 1889 4.2. HBTs with InAs and InGaAs bases . . ....................................................... 1889 4.3. InGaSb/InAlAsSb npn HBTs.............................................................. 1890 4.4. pnp Structures and InP HBTs with GaAsSb bases . .............................................. 1891 5. Summary and outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1891 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1891 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1892 1. Introduction The binary compound semiconductors AlSb, GaSb, InSb, and InAs along with their related alloys are candi- dates for high-speed, low-power electronic devices. Appli- cations could include high-speed analog and digital systems used for data processing, communications, imag- ing, and sensing, particularly in portable equipment such as hand-held devices and satellites. The development of Sb-based transistors for use in low-noise high-frequency ampliﬁers, digital circuits, and mixed-signal circuits could provide the enabling technology needed to address these rapidly expanding needs. In Fig. 1, we show the trend toward higher frequencies and lower power consumption with increasing lattice constant. In recent years, consider- able progress has been made in HEMTs, RTDs, and HBTs in the antimonide–arsenide materials system (lattice constants greater than 6.0 A ˚). In this paper, we will review the progress in the design, growth, fabrication, and performance of these electronic devices. We will dis- cuss our work at NRL as well as major results from other groups. 2. InAs quantum wells and high-electron-mobility transistors 2.1. Background The ﬁrst high electron mobility transistors (HEMTs) were fabricated with GaAs channels and AlGaAs barriers [1]. These devices are also known as modulation-doped ﬁeld eﬀect transistors (MODFETs). In order to achieve higher electron mobility and velocity (translating to higher frequency operation), In was added to the channel. Typical structures have In 0.2 Ga 0.8 As channels that are pseudomor- phically strained to the GaAs lattice constant (PHEMTs). In order to improve performance further, additional In was added to the channel and the barrier material was changed to InAlAs; the larger lattice constants were accommodated by using InP substrates. The logical pro- gression of this trend is to use pure InAs as the channel along with nearly lattice-matched AlSb, AlGaSb, or InAlSb for the conﬁning layer as the arsenides are not suit- able barriers. Advantages of this material system include the high electron mobility (30,000 cm 2 /V s at 300 K) and Ge Si GaAs GaP InP AlSb GaSb InAs AlP AlAs 0.45 0.50 0.60 0.80 1.00 1.30 2.00 5.00 0.8 0.4 1.2 1.6 2.0 2.4 2.8 0 5.4 5.6 5.8 6.0 6.2 6.4 Energy Gap (eV) Wavelength (µm) InGaAs Lattice Constant (Å) InSb GaAs-based HEMTs and HBTs First generation: InP-based HEMTs and HBTs Second generation: Sb-based HEMTs and HBTs Next generation: Fig. 1. Energy gap versus lattice constant, showing the evolution of transistors to larger lattice constants and smaller bandgaps for high-frequency and low-power operation. 1876 B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 velocity (4 ·10 7 cm/s) of InAs [2], and a large conduction band oﬀset between InAs and AlSb (1.35 eV), as shown in Fig. 2 [3–6]. 2.2. InAs/AlSb single quantum wells Most of the recent advances in antimonide-based HEMTs have involved heterostructures grown by molecular beam epitaxy (MBE). Antimonide growth by MBE was ﬁrst reported in the late 1970s [7–9]. The growth of antimonide and mixed antimonide/arsenide structures by MBE has pre- sented some challenges compared to growth of arsenides [10]. For the most part, solutions have been found. Overall, growth of antimonides is simpler than nitrides and does not present the safety issues associated with phosphides. Scale- up to production MBE systems should be feasible. Crucial advances for transistor applications were reported by Prof. KroemerÕs group at the University of California at Santa Barbara (UCSB), beginning in the late 1980s. They grew InAs/AlSb single quantum wells and showed that high mobilities could be achieved by control- ling the interfaces [11,12]. The interface bonds can be either InSb-like or AlAs-like as there is no common cation or anion for a heterojunction between InAs and AlSb. Work by Tuttle et al. showed that the bottom interface must be InSb-like to achieve high mobilities. Bolognesi et al. varied the thickness of the InAs quantum well and achieved high mobilities for 125–200 A ˚[13]. The mobility of thinner wells was suppressed by interface scattering. The decrease in mobility for wells thicker than 200 A ˚presumably resulted from the formation of misﬁt dislocations due to the 1.3% lattice mismatch between the InAs and AlSb. Although it is not the focus of this review, we note that the growth of InAs quantum wells with very high mobilities has lead to many interesting physics studies. For example, see work on ballistic transport phenomena [14,15]. In addi- tion, the Hall eﬀect in these structures is of interest for magnetic sensing [16–18] and magnetoelectronic logic applications [19]. 2.3. Doping of HEMTs Key parameters for HEMT quantum wells are electron mobility and sheet carrier density. Sheet carrier concentra- tions for unintentionally doped InAs/AlSb single quantum wells are known to be a function of the thickness of the upper barrier, the cap material (usually InAs or GaSb), the interface bond type, and the purity of the AlSb. In most cases, the undoped heterostructures are n-type with densi- ties less than or approximately equal to 1 ·10 12 /cm 2 [20]. Higher sheet charge densities are typically desirable for HEMT applications. The arsenic-soak technique, ﬁrst reported by Tuttle et al., is one way to enhance the sheet carrier concentration. After growing an InAs quantum well and an AlSb spacer layer, the AlSb is subjected to an As beam [12]. Apparently, As atoms replace Al atoms, provid- ing extra electrons to the conduction band. Densities of 1–2 ·10 12 /cm 2 are typically achieved. High-speed HEMTs have been fabricated using this technique [21]. Silicon is the most common n-type dopant in III–V MBE systems. However, it is amphoteric in the III–VÕs, producing n-type GaAs, InAs, AlAs, and InSb, but p-type GaSb and AlSb on (1 0 0) surfaces [22]. Sasa et al. achieved sheet carrier concentrations in the 2–4 ·10 12 /cm 2 range by Si planar doping in an 18 A ˚InAs quantum well located 80 A ˚below a 150 A ˚InAs quantum well clad by AlGaSb [23]. The energy levels in the very thin doped well are higher than in the channel due to quantum conﬁnement. Fig. 2. Band alignments for selected binary and alloy semiconductors at 0 K. B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 1877 Hence, electrons will transfer into the channel. This tech- nique has been successfully applied to InAs HEMTs by at least three groups [24–27]. Chalcogens such as S, Se, and Te are in column VI of the periodic table and hence logical candidates for n-type dopants in III–V semiconductors. As discussed by Furuk- awa and Mizuta [28], elemental sources such as Te have a very high vapor pressure and hence are not suitable for III–V MBE. An alternative is to use chalcogenides such as PbSe and GaTe. Wood demonstrated n-type doping of GaAs using PbSe and PbS [29]. Furukawa and Mizuta suc- cessfully used GaTe to make n-type AlGaSb layers in a transistor [28]. Subbanna et al. used PbTe to achieve con- trollable carrier concentrations from 10 16 to 10 18 /cm 3 in a detailed doping study of AlSb and GaSb [30]. Several groups have successfully used GaTe as a dopant source for Sb-based transistors and other structures [31–33]. The relationship between mobility and carrier density is plotted in Fig. 3 for several undoped, Si-doped, and Te- doped HEMT structures grown at NRL. High mobility and low sheet resistance are desired for HEMT applications. Unfortunately, the highest mobilities are achieved for undoped structures with carrier densities that are too low to achieve sheet resistances less than 200 X/h.Fig. 3 demon- strates that sheet resistances near 100 X/hhave been reached for both Si- and Te-doping. We ﬁnd that densities near 3.0 ·10 12 /cm 2 are optimal for achieving both high mobility and low sheet resistance. Note that these high mobilities and low sheet resistances cannot be achieved by GaAs- or InP-based HEMTs. Other groups have achieved results similar to those shown in Fig. 3. To our knowledge, no one has achieved the highest mobilities and high densities simultaneously (e.g. 25,000–30,000 cm 2 /V s and 3·10 12 / cm 2 ). Based upon Shubnikov-de Haas measurements, elec- trons begin to populate a second sub-band in InAs quantum wells at densities of 1.4–2.4 ·10 12 /cm 2 [23,26,34]. Hence, for higher densities we expect intersubband scattering and a reduction in mobility, as shown in Fig. 3. Both Si and Te doping appear to be viable for InAs HEMTs. Advantages of Si include the fact that it is more often available in MBE systems and has a much lower vapor pressure, meaning that source depletion is generally not a problem. A disadvantage of Si is that a decrease in growth temperature may be required between the InAs channel and the doped layer [26]. For Te doping, a con- stant growth temperature can be used. In addition, Te should act as a donor in any AsSb alloy. 2.4. HEMT design and performance The layer structure and calculated band structure for a HEMT with Te delta-doping are illustrated in Fig. 4. Semi-insulating GaAs substrates are typically used for InAs HEMTs as there are no suitable zincblende insulating substrates with lattice constants near 6.1 A ˚. The 1.7 lm Fig. 3. Electron mobility versus sheet density for InAs single quantum wells. The doping technique is indicated next to each point. Ranges for typical GaAs- and InP-based HEMTs are also indicated. AlSb 5 nm InAs 2 nm In0.4Al0.6As 4 nm SI GaAs substrate AlSb 1.7 µm InAs 12 nm AlSb 50 nm Al0.7Ga0.3Sb 300 nm Te delta doping AlSb 1.2 nm -2 -1.5 -1 -0.5 0 0.5 1 1.5 0 20 40 60 80 Energy (eV) Position (nm) AlSb AlSb InAs In0.4Al0.6As Al0.7Ga0.3Sb Te -doping (a) (b) Fig. 4. Heterostructure cross-section (a) and calculated band structure (b) for InAs HEMT with Te delta-doping. 1878 B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 buﬀer layer of AlSb is almost fully relaxed and accommo- dates the 8% lattice mismatch [35]. The 0.3 lm layer of Al 0.70 Ga 0.30 Sb allows for a mesa isolation etch that stops in the AlGaSb and hence does not expose AlSb outside the active region to air or moisture that will cause it to deteriorate [36–38]. The InAlAs layer enhances the insulat- ing properties of the barrier by providing a barrier to holes. In addition, it serves as an etch-stop, enabling a gate recess etch to be employed [21]. Despite the large lattice mismatch and three-dimensional growth, the InAlAs layer does not degrade the electron mobility, presumably because it is at least 50 A ˚above the channel. The AlGaSb and InAlAs lay- ers are now used by at least three groups investigating InAs HEMTs [33,37,39]. The uniformity of transistors across a wafer will be important for circuit applications. Non-uniformities can be introduced in the growth or subsequent processing. Contactless resistivity mapping of an as-grown wafer can easily be done with a Lehighton system as a test of unifor- mity. We have measured non-uniformities as low as 0.7% across a 3-in. wafer (typical values were 1–3%) for Te- delta-doped structures similar to the one in Fig. 4, grown on a 3-in. Riber Compact 21T MBE system [35]. The low-power potential of these materials was demon- strated by the fabrication of InAs HEMTs with a 60 nm gate length [40]. The as-grown heterostructure had a sheet density of 1.6 ·10 12 /cm 2 and a mobility of 21,300 cm 2 /V s. The IVcharacteristics shown in Fig. 5 indicate that the depletion-mode device has a transconductance of 1.1 S/ mm at V DS = 0.35 V and V GS =0.50 V. The increase in output conductance at a drain voltage above 0.3 V is due to the eﬀects of impact ionization in the channel. At this bias condition, the device exhibits an extrinsic f T of 160 GHz. The potential for low-voltage operation is dem- onstrated by the plot of measured f T as a function of drain voltage in Fig. 6.Anf T of 90 GHz is obtained at a drain voltage of only 0.10 V. InP- and GaAs-based HEMTs can- not reach these high frequencies at such low voltages. In the last three years, funding from the DARPA anti- monide-based compound semiconductor (ABCS) program has resulted in substantial advances in InAs HEMT tech- nology. A collaboration between the Naval Research Laboratory and Northrop Grumman Space Technology (NGST) focused on making the devices manufacturable. Signiﬁcant steps in this direction were the elimination of an air bridge used for the device discussed above and the successful fabrication of T-gates. Results from 0.1 lm HEMTs fabricated from a heterostructure with n s = 1.3 ·10 12 /cm 2 and l= 29,000 cm 2 /V s are shown in Figs. 7 and 8 [37,41,42]. The devices displayed high trans- conductance (G m ) at low drain–source voltage (V DS ), and low on-state resistance (R ON ). The average G m peak was 1.05 S/mm and 1.6 S/mm measured at a V DS of 0.2 V and 0.3 V, respectively. The average oﬀ-state reverse gate-drain breakdown voltage (BV GD ) was 1.42 V (measured at a gate current of 1 mA/mm). DC characteristics in Fig. 7 were measured from an 80-lm-gate-width device with typ- ical extrinsic R ON of 0.67 Xmm. (The high peak for DC G m @V DS = 0.4 V is anomalous and related, in part, to impact ionization eﬀects.) These characteristics illustrate the combination of low-drain-voltage operation, low knee voltage, and high transconductance, which are critical parameters for low-power, high-frequency operation. VGS = 0 V Vstep = 0.1 V 0.0 0.1 0.2 0.3 0.4 0.5 Drain Voltage (V) 0 200 400 600 800 Drain Current (mA/mm) Fig. 5. HEMT drain characteristics for an InAs HEMT with a 60 nm gate length, L DS = 1.0 lm, and W G =50lm[40].1999, AVS The Science and Technology Society. VGS = -0.4 V 0.0 0.1 0.2 0.3 0.4 0.5 Drain Voltage (V) 0 40 80 120 160 fT (GHz) Fig. 6. Extrinsic cut-oﬀ frequency, f T , versus drain voltage for a 60 nm InAs-channel HEMT. Note that an f T of 90 GHz is achieved at a bias of only 0.10 V [40].1999, AVS The Science and Technology Society. 0 50 100 150 200 250 300 350 400 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 Gate Voltage (V) fT (GHz) 0 500 1000 1500 2000 2500 DC Transconductance (mS/mm) TOP CURVE: VDS = 0.4 V STEPS OF -0.1 V Fig. 7. Extrinsic cut-oﬀ frequency, f T , and DC transconductance versus gate voltage for 0.1 lm T-gate InAs-channel HEMT [37].IEEE. B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 1879 Small signal RF tests were also performed on-wafer, and the maximum available gain was 10 dB at 100 GHz. The normal slope (6 dB/octave) was used to extrapolate beyond 100 GHz and gave an average peak f T of 153 GHz and 212 GHz at V DS of 0.2 and 0.4 V, and drain current densi- ties of 115 and 340 mA/mm, respectively. These corre- spond to DC power dissipations of 22 and 180 mW/mm. Compared to f T -DC power performance of state-of-the art 0.1-lm gate length InAlAs/InGaAs/InP HEMTs, our AlSb/InAs HEMTs provide equivalent f T at 5–10 times lower power dissipation, as shown in Fig. 8. The use of a T-gate resulted in values of f max greater than f T , with f max = 270 GHz at V DS = 0.4 V. HRL Laboratories recently reported InAs-channel HEMTs with InAlAs barrier layers. The composition and thickness of the In x Al 1x As is not given, but presumably the layers are In-rich and pseudomorphic. For a 70 nm gate length, an intrinsic f T of 308 GHz was achieved at V DS = 0.70 V. The peak intrinsic f max was 110 GHz. The device did not have a T-gate [39]. A Rockwell-UCSB team also achieved excellent low- power performance. They report 0.25-lm T-gate HEMTs fabricated from a heterostructure with n s = 3.7 ·10 12 /cm 2 and l= 19,000 cm 2 /V s [33]. The extrinsic values of both f T and f max are 162 GHz at V DS = 0.35 V, V GS =1.1 V, I d = 6.1 mA, and I g =20.9 lA. In a subsequent report, 0.1-lm T-gates were written on a heterostructure with n s = 2.1 ·10 12 /cm 2 and l= 18,000 cm 2 /V s. These devices reached an f T of 235 GHz at V DS = 0.45 V, and an f max of 235 GHz at V DS = 0.30 V, as shown in the contour plots of Fig. 9 [43]. Note that both f T and f max exceed 100 GHz at a drain bias of only 0.10 V, again highlighting the low- power potential of these transistors. The threshold voltage of the n s = 3.7 ·10 12 /cm 2 Rock- well device discussed above was 1.2 V. We have also observed large negative threshold voltages (V th <1.0 V) for some devices with high sheet charge (3·10 12 /cm 2 ). These high threshold voltages are generally not desirable for low-power circuits. One way to make V th less negative is to decrease the sheet density. The disadvantage to this approach is evident in Fig. 3.At3·10 12 /cm 2 , mobilities near 20,000 cm 2 /V s are possible, yielding sheet resistances near 100 X/h. If the density is reduced to 1 ·10 12 /cm 2 , the mobility may increase to 30,000 cm 2 /V s, but the sheet resistances will increase to 200 X/h. This will mean higher access resistance for the HEMTs. An alternative is to reduce the gate-to-channel spacing. To quantify this approach, we performed Silvaco modeling of V th as a func- tion of sheet density and vertical spacing. We use the dis- tance between the gate (which rests on the InAlAs) and the center of the channel as a parameter. Surface state and quantum eﬀects are not included. The results are shown in Fig. 10. For low densities, the threshold voltage is small (approximately 0.5 V) regardless of the gate– channel separation. For higher densities, there is only a 0 50 100 150 200 250 1 10 100 1000 DC Power Dissipation (mW/mm) Cut-off Frequency, fT (GHz) Increasing VDS VDS = 0.1V 0.2V 0.5V VDS = 0.2V 0.5V 1.0V Fig. 8. Extrinsic cut-oﬀ frequency, f T , versus power dissipation for InAs/ AlSb HEMT (solid lines) at V DS = 0.1, 0.2, and 0.5 V and InAlAs/InGaAs HEMT (dashed lines) at V DS = 0.2, 0.5, and 1.0 V [37].IEEE. 00.1 0.2 0.3 0.4 0.5 0 100 200 300 400 500 600 700 800 Drain–Source Voltage (V) Drain Current (mA/mm) 50 50 50 50 50 100 100 100 100 100 150 150 150 150 150 200 200 200 200 235 fτ(GHz) 0 0.1 0.2 0.3 0.4 0.5 0 100 200 300 400 500 600 700 800 50 50 50 100 100 100 100 100 100 150 150 150 150 150 200 200 200 200 235 Drain–Source Voltage (V) Drain Current (mA/mm) f max (GHz) (a) (b) Fig. 9. Drain current versus drain–source voltage for Rockwell 0.1-lm- gate-length, InAs/AlSb HEMT with contours of f T and f max illustrating high-frequency performance at low power [43].IEEE. 1880 B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 small diﬀerence between the 162 and 214 A ˚lines, but a sub- stantial increase in V th for 300 A ˚. We also include experi- mental points on the plot which generally conﬁrm the predictions. (We have several measured values of 0.4 < V th <0.7 for 1–2 ·10 12 /cm 2 but do not include all of them on the plot for clarity.) It was these results, in part, that led us to select the Te-delta-doped structure shown in Fig. 4. Using structures similar to Fig. 4, we have achieved densities of 1–3 ·10 12 /cm 2 with gate–channel sep- arations of 130–214 A ˚. The reduced spacing is also advan- tageous because it allows good aspect ratios as gate lengths are reduced. 2.5. Variations to standard HEMT design A number of variations to the InAs HEMTs described above have been investigated over the last decade. Several involve changes to the channel. Yang et al. fabricated FETs from a heterostructure with an 80 A ˚InAs/50 A ˚GaSb com- posite channel [44]. By varying the gate voltage, the carriers in the channel can be changed from electrons to holes, resulting in a novel V-shaped transfer characteristic. Boos et al. modiﬁed the conventional InAs HEMTs by adding a42A ˚InAs subchannel, separated from the 100 A ˚InAs channel by 30 A ˚AlSb [45]. The subchannel has a larger bandgap than the channel due to quantization. Hot elec- trons can tunnel from the channel into the subchannel before gaining enough kinetic energy for impact ionization. HEMTs using this design exhibited an intrinsic f T of 250 GHz for a 0.1 lm gate length. A similar approach was taken by Lin et al. who used a composite channel of InAs and InAsP (InAlAs) with the InAsP (InAlAs) having a larger bandgap [46,47]. The resulting HEMTs exhibited increased breakdown voltage compared to InAs-channel devices. A ﬁnal variation to the channel is to incorporate alloys of InAsSb. Using digital alloys of InAsSb, we showed that the band alignment between the InAs 1x Sb x and AlSb changes to type-I for x> 0.15 [48]. The type-I alignment should enable better conﬁnement of holes gener- ated thermally or as a result of impact ionization in the channel. We fabricated 0.1 lm HEMTs with InAs 0.8 Sb 0.2 channels and achieved intrinsic values of f T = 180 GHz and f max = 120 GHz, and a voltage gain of 9, despite the fact that the mobility was only 13,400 cm 2 /V s [49]. More recently, we have improved the mobility to 22,000 cm 2 / V s using random alloys of InAs 0.7 Sb 0.3 and barriers of In 0.2 Al 0.8 Sb. Transconductances of 1.35 S/mm were mea- sured at V DS = 0.3 V for a 0.2 lm gate length [50,51]. Another variation we investigated included a 100–200 A ˚ layer of p-type GaSb (doped in the mid 10 17 /cm 3 by Si or Be) 500 A ˚below the InAs channel [21]. The intention of this layer is to drain a portion of the impact-ionization-gen- erated holes back to the source contact rather than having them remain in the AlSb buﬀer layer where they are likely to cause deleterious trapping eﬀects or be collected at the gate contact and thereby increase the gate leakage current. A related technique is to use an epitaxial back-gate placed below the quantum well to remove holes [52]. Another method to minimize impact ionization eﬀects is the use of dual gates. The use of a dual gate for InAs HEMTs was proposed in 1990 [53] and demonstrated by two groups in 1996 [54,55]. This design enables the reduc- tion of leakage current by using a second gate to modify the E-ﬁeld under the ﬁrst gate. A disadvantage is the increased process complexity required to achieve low access resistance with the dual-gate design. As mentioned earlier, even undoped InAs quantum wells usually have 10 12 /cm 2 electrons in the channel. Hence, InAs HEMTs normally operate as depletion-mode devices. Zhao et al. incorporated Be delta-doping in the upper barrier to achieve n-channel enhancement-mode transistors [56]. Recently, other groups have also been working toward this goal [39,57,58]. The typical InAs HEMT structures include a thin (2 nm) fully-depleted InAs cap which is removed to allow a recessed gate on the InAlAs barrier. An alternative, com- monly used in InP-based HEMTs, is to grow a thicker n + cap layer. This can result in low access resistance and low threshold voltage. In a recent report, HEMTs were fabri- cated on a heterostructure with a 20 nm n + -InAs cap above an InAlAs/AlSb composite barrier [59]. At NRL, we are investigating structures with a 20 nm n + -InAs cap above an InAlSb barrier [60]. 2.6. Microwave and low-frequency noise As discussed earlier, several groups have now demon- strated low-power InAs HEMTs with values of f T and f max exceeding 100 GHz. If this technology is to become viable, however, the microwave noise must be low. As shown in Fig. 11,F min > 1 dB was measured for frequencies of 2– 20 GHz in the ﬁrst noise measurements reported in 1997 [61]. These relatively large noise ﬁgures were caused by the high gate leakage currents. Simulations indicated that a reduction in gate leakage current should result in much lower noise ﬁgures (e.g. 0.3 dB at 4 GHz). Devices recently fabricated by Rockwell [43,62] and NGST/NRL [37,63] 0 -0.5 -1 -1.5 -2 0 0.5 1 1.5 2 2.5 3 3.5 4 Sheet Char g e (1012cm-2) Threshold Voltage (V) 300Å 214Å 162Å 265Å 214Å 315Å 164Å 250Å 264Å 240Å 300Å Fig. 10. Calculated threshold voltage as a function of sheet charge at three diﬀerent values of gate-to-center-of-channel separation. Experimental data points are also shown. B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 1881 exhibited lower gate leakage current and a substantial reduction in F min , as shown in Fig. 11. These results are very encouraging: they demonstrate that InAs HEMTs have attained microwave noise performance comparable to GaAs- and InP-based HEMTs. The low-frequency noise of InAs- and InAsSb-channel HEMTs is also of considerable interest for two primary reasons. First, this noise is a very sensitive indicator of material quality, particularly at the interfaces, and there- fore can be used to ﬁne-tune material composition and growth parameters. Secondly, since this noise is upcon- verted in non-linear HEMT circuits, it has a direct bearing on the phase noise of microwave oscillators and phase jitter of high-speed digital circuits. At this time, measurements of the low-frequency noise characteristics of AlSb/InAs/AlSb [64,65] and AlSb/InAsSb/AlSb [66] HEMT structures have been performed. Typical devices whose noise spectra were studied consisted of small HEMT structures, 30 lm wide, with a source–drain length of 3 lm, and gate length of 0.1 lm. The various layer structures examined included InAs channels, digital alloy superlattice channels of InAs/ InSb, and random alloy channels of InAsSb. Both undoped and doped devices were included in the study. The mea- surements were usually done open-channel to avoid gate leakage current whose associated shot noise could aﬀect the results. Although considerable variations among the noise char- acteristics were observed, several points can be made to summarize the results of the measurements. The low-fre- quency noise in the HEMTs consists of the superposition of 1/fnoise and a local-level component which contributes noise in the temperature range between 150 K and 300 K. An Arrhenius plot of the level yields an activation energy of about 0.38 eV. The value of the Hooge parameter, a H , was found to range between 10 3 and 10 2 . No correlation between the channel mobility and the Hooge parameter was found in the samples measured. A correlation between the noise and the channel strain, however, was suggested by the results. Speciﬁcally, the devices with InAs 0.8 Sb 0.2 chan- nels, which are unstrained, had the lowest noise, while those with InAs-channels (in tension) and those with InAs 0.7 Sb 0.3 (in compression) were noisier. It should be noted that no eﬀort has been made at this point to optimize the layer structure, the growth parame- ters, or the processing procedure in order to minimize the low-frequency noise. Although the Hooge parameter in these devices is about one order of magnitude higher than in the best InP- or GaAs-based HEMTs, the values of a H appear reasonable for a relatively immature technology with a dislocation density of 10 8 –10 9 cm 2 due to lattice mismatch with respect to the substrate [67,68]. It should also be noted that in all the devices measured at NRL, the noise increases greatly with illumination at low temper- atures, and displays a strong Lorentzian spectrum. This noise appears to involve a deep level in the AlSb which can host the photo-excited electrons and thereby also con- tribute to negative photoconductivity. 2.7. Radiation eﬀects The low power consumption of InAs HEMTs makes them candidates for applications in space where toler- ance to radiation eﬀects is clearly an important issue. Pre- liminary measurements of laser-induced single-event eﬀect (SEE) characteristics have been performed [69]. The results indicate SEE performance will be similar to that observed previously for InP HEMTs and GaAs FETs under similar excitation conditions. A recent study used proton bom- bardment to compare the radiation-induced change in drain current for several HEMT material systems [70]. InAs HEMTs were not included in the study, but the con- duction band oﬀset between the channel and barrier mate- rials was found to be a key parameter, with larger oﬀsets resulting in smaller changes in drain current. Hence, the 1.35 eV conduction band oﬀset in InAs/AlSb HEMTs should result in excellent tolerance to radiation. Recent experimental results conﬁrm this, with the radiation- induced decrease in HEMT drain currents a factor of 150 lower than for typical GaAs/AlGaAs HEMTs [71]. 2.8. Circuits The ﬁrst InAs HEMT circuits were reported in 2003– 2005. HRL reported an 18-transistor inverter-buﬀer [39]. The circuit was not designed for depletion-mode devices but the DC transfer curves exhibit the correct functionality. Using a conservative design, NGST/NRL demonstrated an X-band low-noise ampliﬁer (LNA) with greater than 7 dB/ stage peak gain between 12 and 14 GHz and 6 mW/stage DC power dissipation [37]. Rockwell fabricated low-power Ka-band LNAs with a noise ﬁgure of 1.46 dB at 35 GHz (see Fig. 11) and an associated gain of 22 dB.[72] Most Fig. 11. Noise ﬁgure, F min , as a function of frequency for InAs-channel HEMTs [37,61–63,72]. 1882 B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 recently, a W-band LNA was reported by NGST/NRL. The 3-stage ampliﬁer has a noise ﬁgure of 5.4 dB with an associated gain of 11.1 dB at a total chip dissipation of only 1.8 mW at 94 GHz [63]. The dc power consumption for both the Ka and W-band LNAs was less than one-tenth (one-third) the power for comparable GaAs(InP)-based circuits. 2.9. InSb-channel HEMTs InSb has a room-temperature electron mobility of 78,000 cm 2 /V s, the highest of any semiconductor. In addi- tion, its electron saturation velocity is reported to be greater than 5 ·10 7 cm/s, compared to 4 ·10 7 and 1·10 7 cm/s for InAs and GaAs, respectively [73]. In prin- ciple, transistors with InSb channels could reach higher fre- quencies than the InAs-channel HEMTs. Quantum wells can be formed with InSb wells and InAlSb barriers. Room-temperature mobilities as high as 40,000 cm 2 /V s have been achieved [74]. The mobility decreases with increasing carrier density, resulting in sheet resistances greater than 200 X/h. The high mobility of InSb should enable low-voltage operation. The high intrinsic carrier concentration (n i 10 16 /cm 3 ) and thermal generation rate result in high gate leakage currents. This problem has been mitigated by the use of a carrier extraction technique invented at QinetiQ Corporation [73]. The basic device structure is shown in Fig. 12. The contact to the p + -InAlSb layer allows the minority carrier density to be reduced by many orders of magnitude. For a gate length of 0.2 lm, an f T of 150 GHz was achieved at V DS = 0.5 V [75]. 3. Resonant tunneling diodes 3.1. Background Resonant tunneling diodes (RTDs) were ﬁrst fabricated in the GaAs- and InP-based material systems and later in the antimonide system, similar to the situation for HEMTs [76]. In 1988, Luo et al. fabricated RTDs with InAs contact layers, AlSb barriers, and an InAs well [77]. The band dia- gram is shown in Fig. 13a. Electrons tunnel via conduction band states in the well in this type-I structure. The follow- ing year, So ¨derstro ¨m et al. reported on the ﬁrst resonant interband tunneling diode (RITD) with the InAs well in the RTD structure replaced by GaSb [78]. As shown in Fig. 13b, the band alignment is type II, with the valence band edge of the GaSb above the conduction band of the InAs. Interband tunneling through the structure occurs when electrons from the InAs conduction band tunnel through the AlSb barrier, into the GaSb valence band, through the second barrier, and into the InAs conduction band. To minimize power dissipation in circuits using res- onant tunneling devices it is desirable to have the current peak occur at a low voltage. SchulmanÕs calculations shown in Fig. 14 [79] illustrate the diﬀerence between the peak positions of Sb-based RITDs and RTDs. Note that the current peaks at a bias near 100 mV (deﬁned as V p ) for the RITD compared to 200 mV for the RTD. The liter- ature for Sb-based RITDs shows a wide range of V p values with some as low as 100 mV. The Sb-based RITD peak voltages are also much smaller than the bias voltages of 1–1.5 V that are generally required to reach I p in GaAs- and InP-based RTDs with the smallest being near 0.4 V [80]. While the RTD and RITD structures in Fig. 13 have semi-insulating GaAs substrate InSb well InAlSb buffer Source Gate Drain Extraction (optional) p+ InAlSb buffer InAlSb barrier InAlSb barrier Modulation doping (optional) Doped channel (optional) Fig. 12. Cross-section of InSb-channel HEMT with additional contact for carrier extraction [75]. InAs InAs AlSb AlSb ER tt InAs (a) InAs InAs AlSb AlSb GaSb ER tt (b) Fig. 13. Conceptual band structure for antimonide-based (a) resonant tunneling diode (RTD) and (b) resonant interband tunneling diode (RITD). B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 1883 been the most often studied structures, the unusual align- ment of the valence and conduction bands and the large diﬀerences in the bandgaps allow a wide variety of struc- tures to be made. Many of them have been reported in the literature [81–91]. 3.2. Performance: peak current density and peak-to-valley ratio We show typical IVdata from one of our RITDs in Fig. 15. The structure was grown on an InAs substrate and was symmetric, consisting of a 1.0 lm InAs (n3· 10 18 /cm 3 ) buﬀer layer, 300 A ˚InAs (n1·10 17 /cm 3 ), 120 A ˚undoped InAs, 15 A ˚AlSb, 80 A ˚GaSb, 15 A ˚AlSb, 120 A ˚undoped InAs, 300 A ˚InAs (n1·10 17 /cm 3 ), and 0.2 lm InAs (n3·10 18 /cm 3 ). The data shown here are for a 3-lm-diameter circular mesa. For positive biases, the ﬁgures of merit are: peak current density, I p = 1.2 · 10 4 A/cm 2 and peak-to-valley ratio (PVR) = 16. For nega- tive biases, I p =1.3 ·10 4 A/cm 2 and PVR = 15. Large values of I p are required for high-speed operation in order to rapidly charge and discharge internal and circuit capa- citances [92,93]. The low valley currents are also desirable for low power dissipation. The peak and valley currents in Fig. 15 are found at the small biases of 120 mV and 300 mV, respectively. This is an important feature of this material system as these low voltages oﬀer the possibility of building low-power-dissipation electronic devices. We have measured devices with peak currents at voltages as low as 65 mV. Calculating the tunneling current in the InAs/AlSb/ GaSb system is very diﬃcult as it requires using at least 8 bands to calculate the wave functions in the various layers for many diﬀerent values of the in-plane and out-of-plane momentum for each bias. The solution is further compli- cated by the need to satisfy both the Schro ¨dinger and Pois- son equations. In addition to ideal elastic tunneling, the current is determined by a number of phenomena such as phonon interactions [94], interface roughness scattering [95], and impurity scattering [96]. Most of these interac- tions have been studied in detail in the GaAs/AlGaAs sys- tem but not the interband system discussed here. A starting point for an investigation of resonant inter- band tunneling is determining the dependence of I p and the PVR on the barrier thickness. An exponential depen- dence of I p on the AlSb barrier thickness is anticipated because the transmission probability for an electron to cross a barrier increases exponentially with decreasing bar- rier thickness. In Fig. 16a, we plot I p versus barrier thick- ness for a set of six samples we grew under nominally identical conditions. Data from other groups is also shown and is in good agreement with our results. The exception is the data point from the ﬁrst reported RITD; presumably growth procedures were not yet optimized. Fig. 16a indi- cates that the current increases almost exponentially with decreasing AlSb barrier thickness for thicknesses greater than 10 A ˚, and somewhat slower for smaller thicknesses. Kitabayashi et al. have reported on the dependence of the tunneling current on the width of the GaSb well and AlSb barrier for barrier widths down to zero [97]. They analyze the diﬀerence between the peak current and the val- ley current and explain their observations in terms of the location of the resonance level in the GaSb well relative to the conduction band edge in the InAs, and the width of the resonance. Their qualitative discussion is based on several assumptions including one that the light-hole band is the dominant band in the GaSb well. They assume that the resonance level moves towards the GaSb valence band edge as the well widens, and that it depends on the barrier thickness. For thin barriers there is a large overlap between the InAs conduction band and GaSb valence band states leading to a smaller eﬀective mass for thin barriers com- pared to thick barriers. The eﬀective mass contributes to determining the location of the resonance level in the well. The dependence of the PVR on the barrier thickness has also been examined. In Fig. 16b we show the PVR values corresponding to the data in Fig. 16a and observe a maxi- mum of 20 at an AlSb barrier thickness of 21 A ˚. There are many possible contributing factors in determining the Fig. 14. Calculated IVcharacteristics for antimonide-based RTDs and RITDs. Note the small bias voltages [79]. Fig. 15. Current–voltage data for an NRL InAs/AlSb/GaSb/AlSb/InAs RITD with 15 A ˚barriers. 1884 B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 valley current, and the diﬀerence in the thickness dependen- cies of I p and PVR suggests that diﬀerent eﬀects dominate at diﬀerent thickness ranges. Shen measured the tempera- ture dependence of the valley current at several diﬀerent voltages and proposed that ﬁeld-assisted thermionic hole emission dominates at room temperature and above, and that Fowler-Nordheim hole tunneling is dominant at low temperature and high ﬁelds [98]. More recently, Xu et al. examined the temperature dependence for several devices with barriers from 6 to 20 A ˚thick [99]. The valley current increased exponentially in all cases but the increase was fas- ter for the thicker barriers. This led to the conclusion that the predominance of thermal emission over hole tunneling increases with increasing barrier thickness. Tehrani et al. reported some success in reducing valley currents by add- ing a monolayer of AlAs to the barriers to increase the bar- rier height [100]. This has limited usefulness as it also reduces the peak current and there is a limit to how thin the barrier can be. Defects have also been shown to add to the valley cur- rent. Shiralagi et al. noted this in a study of the thickness of the InAs buﬀer needed when using a GaAs substrate to obtain the same PVR as found for an InAs substrate [101]. Magno et al. reported observing features in AFM measurements on RITDs with small PVR that were not present in the topography of diodes with larger PVR [102]. These features are believed to be associated with defects that propagate from the substrate through the RITD layers. RITDs have been exposed to 2 MeV protons in order to study the incremental changes in the peak and valley currents due to defect-assisted processes. Diodes with a thick, 13 ML barrier were more sensitive than those with 5 ML barriers. In both cases the valley current increased faster than the peak current [103]. Ternary alloys have been used in variations of the stan- dard RITD structure shown in Fig. 13 to increase the peak current and PVR. Schulman et al. modiﬁed the RITD structure by replacing the GaSb well with GaAs x Sb 1x [104]. The strain should move the heavy-hole band to an energy below the light-hole band. This is expected to reduce the high-bias valley current through the heavy-hole resonance resulting in an improvement in the PVR. Results for several samples with values of xfrom 0 to 0.3 revealed a maximum PVR = 11 at x= 0.1 compared to PVR = 8 at x= 0 and PVR = 3 for larger x. Unfortunately, the peak current dropped signiﬁcantly from a high at x=0. At NRL, we replaced AlSb barriers with AlGaSb and InAlSb in RITD structures. The smaller bandgaps resulted in increases in I p by as much as a factor of three [105]. PVR values generally decreased for the alloy barriers but remained above 10. An additional advantage of these alloys is that their oxidation rates are substantially lower than pure AlSb which should improve device reliability [38]. The peak current density and PVR for InAs/AlSb RTDs (see Fig. 13a) from several groups are shown in Fig. 17. Compared to the RITDs, higher peak currents have been achieved for antimonide RTDs, with several reports of I p greater than 1 ·10 5 A/cm 2 [82,106,107]. The PVRs for the RTDs are smaller than those for the RITDs, and the PVR for the RTD data of So ¨derstro ¨m-1990 increases monotonically with thickness unlike the RITD that has a maximum [108]. When an RITD is biased beyond resonance, the electrons must tunnel through the bandgap of the GaSb well in addition to the AlSb barriers. Hence, tunneling contributions to valley currents are min- imized, resulting in higher PVRs for RITDs [78]. The values of I p and PVR for Sb-based RTDs and RITDs are clearly adequate for many potential applica- tions. There have been reports of RTDs using InGaAs alloys for the electrodes and wells along with strained AlAs barriers or unstrained InAlAsSb barriers on InP substrates. PVRs near 50:1 were obtained but with relatively low peak currents of 5.8 ·10 3 A/cm 2 [109,110]. The peak voltages reported for both these cases were over 750 mV. An Fig. 16. Peak current density (a) and peak-to-valley ratio (b) as a function of AlSb barrier thickness for symmetric antimonide-based RITDs reported by several groups [78,97,101,102,104,112]. B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 1885 I p = 4.1 ·10 5 A/cm 2 has been reported for InP-based RTD with a strained In 0.8 Ga 0.2 As well and strained AlAs barri- ers. The PVR = 3.8 was reported along with a peak voltage of 680 mV [80]. 3.3. Bias voltages and hysteresis Some care needs to be used in comparing the peak volt- ages reported in the literature, as all devices will contain some series resistance due to the geometry and quality of the ohmic contacts. The peaks may also be shifted by a voltage drop across depletion regions in the electrodes adjacent to the barriers. The size of the variations in V p that can result from series resistance eﬀects are illustrated in Fig. 18 [111] by the IVcharacteristics for RITDs on the same wafer with mesa diameters of 3 lmand95lm. Here the small device has V p 90 mV while it is shifted to 400 mV for the large device. The peak shift can be understood by recognizing that the applied bias voltage is given by the sum of the voltage drop across the RITD and that across the series resistance. The series resistance is not expected to scale with diode area as there are many contributing factors in addition to mesa diameter. For the 3 lm device, the intrinsic RITD resistance dominates, but for the 95 lm device the series resistance is signiﬁcant, causing an increase in V p . This data also illustrates the large hysteresis in the negative resistance region that can result from parasitic series resistance. We have consistently obtained values of V p near 100 mV for RITDs by using small-diameter mesas with short paths through the InAs to the ohmic contacts as well as short wires to the electron- ics. Other groups have also reported values near 100 mV [78,97,112]. For Sb-RTDs, values of V p as low as 200 mV have been reported [113] while higher values are also found in the literature, presumably due to series resistance eﬀects. The hysteresis exhibited by the RITD in Fig. 18bis believed to be largely an extrinsic eﬀect since it disappears for the small devices on the same wafer. Hysteresis and peak shifts may be due in part to the storage of charge in the well which is an intrinsic phenomena. Charge stor- age may also lead to bistabilities that would result in Fig. 18. IVcharacteristics of RITD devices: (a) 3-lm-diameter mesa and (b) 95-lm-diameter mesa. The solid and dashed lines in (b) represent voltage sweeps of diﬀerent directions, as indicated by arrows [111]. IEEE. Fig. 17. Peak current density (a) and peak-to-valley ratio (b) as a function of AlSb barrier thickness for symmetric antimonide-based RTDs reported by several groups [77,106–108,113]. 1886 B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 high-frequency oscillations. Chow and Schulman [114] found that RTDs with AlGaSb barriers and InAs wells and contact layers exhibit an intrinsic bistability. Jimenez et al. also demonstrated intrinsic bistability for RTDs with AlSb barriers and GaSb wells and contact layers [115]. Diode bistability could be useful in the design of high-fre- quency oscillators and bistable logic circuits. 3.4. Substrates, defects, and interfaces As discussed in Section 2, there are no suitable semi- insulating substrates with a lattice constant near 6.1 A ˚. Discrete RTDs and RITDs can be grown on conducting InAs or GaSb substrates, but circuit applications will gen- erally require SI-GaAs or -InP substrates or wafer transfer techniques. Brown et al. compared InAs/AlSb RTDs grown on GaAs and InAs substrates and found little diﬀer- ence in the PVR values, despite the high density of disloca- tions in the GaAs sample [116]. For RITDs, we have generally found that some of the diodes on GaAs sub- strates have PVR values equal to devices on InAs sub- strates, but the uniformity is not as good. AFM imaging reveals that devices which contain a defect have substan- tially lower PVR values [102]. Shiralagi et al. looked at RITD performance on GaAs substrates as a function of the InAs buﬀer layer thickness. PVR values increased as the InAs thickness increased, reaching values of 14 for 1lm buﬀers, compared to 16 for InAs substrates [101]. If antimonide R(I)TDs are to be used in circuit applica- tions, the uniformity and reproducibility of I p will be important. State-of-the-art production MBE systems can achieve thickness uniformities of 1% across large-diame- ter substrates. Hence, we do not expect signiﬁcant variation in I p as a result of thickness variation across wafers. The presence of defects, especially in the case of lattice-mis- matched growth, could be an issue. We have found that defects primarily inﬂuence the PVR rather than I p . In most applications, PVR uniformity will be less important than I p uniformity. In one case we found that the PVRs across an RITD wafer varied from 18 (no defect) to 9 (defect present) [102]. If, for example, the minimum PVR required for a cir- cuit was 5, then this defect-related non-uniformity would be acceptable. A more serious issue may be run-to-run reproducibility. As mentioned above, I p is a strong func- tion of AlSb barrier thickness. To obtain high values of I p (>1 ·10 4 A/cm 2 ), AlSb barriers less than 15 A ˚thick are required for RITDs. Hence, variations of 1 A ˚will cause signiﬁcant changes in I p (see Fig. 16a), and the quality of the interfaces is expected to play an important role. We grew sets of RITDs in which the growth temperature was varied from 350 to 500 C with other parameters ﬁxed. Remarkably, NDR behavior was observed across this entire range, with values of I p and I v varying by less than 20% (for example, the data in Fig. 15 are from a device grown at 350 C). This suggests that precise control of growth temperature will not be necessary for good repro- ducibility. The variations that we did observe can be explained by smoother interfaces at higher growth temper- atures, resulting in barriers that are eﬀectively thicker, yielding lower I p values [102]. As discussed in Section 2, either InSb- or AlAs-like bonds can form the interface between InAs and AlSb. Unlike the situation with the HEMTs, however, RITDs can be produced using either bond type [100,101]. In fact, PVR values as high as 30:1 for negative biases were achieved using AlAs interfaces [100]. (These structures were intentionally asymmetric, with much larger PVR but lower I p for negative biases than for positive bias. Hence, we do not include them in Fig. 16.) Using STM, we found that a quarter monolayer of roughness is introduced at the InAs/ AlSb interface due to a change in surface reconstructions when growing InSb interface bonds. This roughness can be eliminated by intentionally depositing 1.25 ML In (rather than 1.0 ML) at the interface [117]. We grew RITDs in which only the InSb-like interface was varied (1.0 and 1.25 ML In). The room-temperature IVcharacteristics were very similar. These samples were also characterized by Shubnikov-de Haas oscillations at 1.7 K. The sample with 1.25 ML In showed better-deﬁned oscillations, con- ﬁrming that the interface was indeed smoother [118]. Some potential applications for antimonide R(I)TDs require very high densities of devices and hence small-diam- eter diodes. Nomoto et al. fabricated Sb-RTDs with diame- ters ranging from 100 lmto20nm[119,120]. They observed a reduction in PVR for devices smaller than 1 lm and attrib- uted it to a contribution of the surface current to the valley current. Even at 20 nm, NDR behavior was observed. In contrast, the contact layers are pinched oﬀ by surface deple- tion in GaAs RTDs for diameters less than 100 nm [121].In related work, Shiralagi et al. used epitaxial regrowth on patterned substrates to deﬁne sub-micron Sb-RITDs with- out using any ﬁne-line lithography [122]. 3.5. Radiation eﬀects As mentioned in Section 2, radiation tolerance may be important for Sb-based devices. We reported preliminary measurements for InAs/AlSb/GaSb RITDs [103]. The diodes were able to sustain signiﬁcant proton irradiation before the IVcharacteristics deteriorated. As mentioned earlier, diodes with 5 ML barriers were less sensitive than ones with 13 ML barriers. This is fortunate because the thinner barriers will be required to achieve the large current densities needed for high-frequency operation. Two diﬀer- ences have been noted in the response of an InP-based RTD and an RITD to radiation. The peak current for the InP-based diode decreased with ﬂuence while it increased for the RITD, and an InP-based RTD was more sensitive than the 5-ML-barrier RITD. Additional work is necessary to fully understand these diﬀerences. We have also investigated laser-induced single-event eﬀect character- istics [111]. The measurements reveal complex behavior depending on whether the device is biased at a voltage below or above the voltage of the current peak. The B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 1887 response also occurs with two decay times: a nanosecond and 100 picoseconds for an RITD with a peak current den- sity of 10 3 A/cm 2 . A qualitative description based on the diﬀerent time scales for the decay of electrons and holes stored in the well has been proposed. 3.6. Applications Several applications for Sb-R(I)TDs have been pro- posed and demonstrated. Chow and Williamson et al. com- bined RITDs and Schottky diodes to produce a complete set of logic gates with a maximum operating frequency greater than 12 GHz [112,123]. Shen et al. used RITDs to fabricate static random-access memories and demonstrate bistability and switching [124]. They combined RITDs with InGaAs-channel FETs to demonstrate an exclusive-NOR logic gate [125]. Fay et al. also combined Sb-RITDs with an InGaAs-channel FET and demonstrated an integrated circuit based upon the monostable/bistable logic element (MOBILE) [126]. At NRL, we integrated Sb-RITDs with Sb-HEMTs, as shown in Fig. 19. Both the transistor and diode exhibited good performance (HEMT f T = 220 GHz; RITD I p = 1.4 ·10 4 A/cm 2 and PVR = 11:1) [127].Cir- cuits have not been fabricated but modeling suggests that MOBILE circuits with very low power dissipation should be feasible [128]. As discussed in Section 2, Sb-HEMTs exhibit lower power dissipation than InP-based HEMTs, due to operation at lower bias. We have also combined RITDs with giant magnetoresistance elements and demon- strated monostable–bistable transition logic elements [129]. The fast response times were discussed in Section 3.5. Since RITDs with current densities near 10 5 A/cm 2 have been reported, and the frequency scales with current density, RITDs can be expected to operate at frequencies above 100 GHz. In a landmark study, Brown et al. reported an InAs/AlSb RTD operating at 712 GHz [130]. The ﬁnal application we will discuss in this section is the mm-wave backward diode. The basic structure consists of a single AlSb barrier, separating n-InAs and p-GaSb, as shown in Fig. 20a. Luo et al. [131] and Chen et al. [132] demonstrated negative resistance from this structure. Schulman and co-workers applied these diodes to mm- wave detection, translating low-level rf power into dc 1000 Å n+InAs 300 Å n-InAs 120 Å InAs 15 Å AlSb barrier 80 Å GaSb well 15 Å AlSb barrier 120 Å InAs 300 Å n-InAs 1000 Å n+InAs Etch Stop 20-50 Å InAs 40 Å In0.4Al0.6As 12 Å AlSb 12 Å n+InAs(Si) doping 125 Å AlSb 100 Å InAs channel 30 Å AlSb 42 Å InAs sub-channel 500 Å AlSb 100 Å p+GaSb(Si) 2 m AlSb buffer SI GaAs (001) substrate Source Gate Drain Fig. 19. Cross-section for antimonide-based RITD and HEMT grown in a single heterostructure at NRL [127].2000, AVS The Science and Technology Society. Fig. 20. Band diagram and IVcharacteristics from HRL backward- diode for mm-wave detection [133,134].IEEE. 1888 B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 voltage or current with extreme linearity and low noise [133–135].InFig. 20b, we show their IVcharacteristic for a 1.5 ·1.5 lm 2 detector. The large curvature at zero bias translates into good responsivity at a frequency of 95 GHz. Unlike other Sb-based devices discussed in this review, the mm-wave detector is already used in a commer- cial product. 4. Heterojunction bipolar transistors 4.1. Background and early work The design and growth of an HBT is generally more challenging than a HEMT or R(I)TD because of the need to obtain the appropriate conduction- and valence-band oﬀsets at the emitter–base and base–collector heterojunc- tions. High current gains require low carrier recombination rates in the base and at the emitter–base junction, thus low defect concentrations are desirable. High mobilities are desirable to minimize the transit time across the base and collector depletion regions, enabling high-frequency opera- tion. High mobilities and low resistance ohmic contacts also help to minimize parasitic resistances that slow devices through large RC time constants. High-frequency opera- tion requires large collector currents. Hence, power dissipa- tion becomes a problem that can be handled by using high-thermal-conductivity materials and devices that oper- ate at low voltage. To attain high performance requires making wise choices particularly in the choice of materials given that optimum device geometry may be obtained in any material system. These facts, combined with the rela- tive immaturity of the antimonide-based materials systems, resulted in relatively little work on antimonide-based/nar- row-bandgap HBTs until recently. This system has advan- tages in that materials such as InAs and InGaSb have narrow bandgaps and high electron and hole mobilities that make them attractive for use as the base. The small bandgaps should lead to low power dissipation through use of small emitter–base and emitter–collector voltages, and the high mobilities should lead to high-frequency oper- ation. The choice of semiconductors for the emitter and collector with suitable bandgaps and band oﬀsets becomes a major consideration. They should be closely lattice matched to the base, and have a large valence band oﬀset relative to the base. While there is data in the literature on the bandgaps and band oﬀsets of some of the relevant materials there are many uncertainties in the details. There are a number of publications listing interpolation routines and bowing parameter data that are useful as a starting point. The lack of a commercial semi-insulating substrate with a lattice constant near that of InAs and GaSb is another diﬃculty. Vergurlekar et al. published their results on the develop- ment of a hot-electron AlSb/InAs bipolar transistor in 1990. They showed that the large conduction band discon- tinuity between InAs and AlSb resulted in impact ioniza- tion when electrons were injected into a p-InAs base across the AlSb/InAs heterojunction [136]. Two years later, Pekarik et al. demonstrated the ﬁrst AlSb–InAs–AlSb pnp HBT. The devices exhibited current gain but were limited by extremely large base currents, apparently due to exten- sive interface recombination [137]. Dodd et al. fabricated npn InAs bipolar transistors on InP in an attempt to achieve pseudo-HBT performance due to bandgap narrow- ing. Current gains of 30 were achieved at room temperature but junction leakage currents were very high. The dc per- formance was improved when the structure was grown on an InAs substrate [138]. Transistors grown on a conduc- tive substrate, however, cannot be used at microwave fre- quencies. For this reason, Moran et al. used wafer bonding to transfer an antimonide-based npn HBT from a GaSb substrate to an insulating sapphire substrate [139]. The transistors exhibited a dc current gain of 5. 4.2. HBTs with InAs and InGaAs bases One approach to narrow-gap HBTs is to use InAs as the base with alloys as the emitter and/or collector. Averett and co-workers fabricated npn HBTs with an InAs base and collector and InAlAs or InAsP emitters [140–142]. Thomas et al. at HRL used InAsP as both the emitter and collector [143].InFig. 21, we plot the collector current density, J C , as a function of the emitter–base voltage, V BE , for several materials systems [144,145]. As expected, as the bandgap of the base decreases, the emitter–base voltage for a given collector current also decreases. The HRL devices demonstrated low-power operation, with gain for V BE between 0.1 and 0.3 V. When transferred to a sapphire sub- strate, the HBTs exhibited excellent microwave perfor- mance with an f T of 181 GHz. Divide-by-16 circuits consisting of 62 transistors were demonstrated and exhib- ited a maximum operating frequency of 10 GHz [143]. Conventional HBTs lattice-matched to SI-InP sub- strates use In 0.53 Ga 0.47 As as the base with either InP or In 0.52 Al 0.48 As for the emitter and InP or In 0.53 Ga 0.47 As as the collector. A narrow-gap alternative is to use InAlAs 0.2 0.4 0.6 0.8 VBE (V) 10-1 100 101 102 103 104 105 InAsP/InAs/ InAsP (HRL) InAlAsP/InGaAs/ InAlAs (NGST) SiGe InP InAlAsSb/InGaSb/ InAlAsSb, 6.2 Å simulation (NRL) InAlAsSb/InGaSb/ InAlAsSb, 6.3 Å simulation (NRL) Current Density (A/cm2) Fig. 21. Collector current density versus base-emitter voltage for npn HBTs [143–146,149]. B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 1889 and InGaAs with indium concentrations greater than 80% in all three layers. Monier et al. at NGST demonstrated npn HBTs using an In 0.86 Al 0.14 As/In 0.86 Ga 0.14 As/In 0.86 A- l 0.14 As structure with a 6.0 A ˚lattice constant [146]. The devices were grown on InP substrates using graded InAlAs buﬀer layers [147]. Gain was achieved with V BE between 0.2 and 0.4 V, as shown in Fig. 21. Microwave measurements yielded a peak cut-oﬀ frequency of 170 GHz. Initial test cir- cuits include 28 GHz dividers that dissipate half the power of InP-based circuits [148]. 4.3. InGaSb/InAlAsSb npn HBTs At NRL, we surveyed materials and band oﬀsets and selected a material system with a lattice constant near 6.2 A ˚[149]. The willingness to try working with a large lat- tice mismatch was motivated in part by the success in the InP community in developing metamorphic growth tech- niques allowing the use of GaAs substrates in the growth of HBTs and HEMTs lattice matched to InP. Past success at NRL in growing InAs/AlSb HEMTs on GaAs also was considered. For developmental purposes commercial GaSb substrates were chosen to minimize the lattice mismatch rather than the preferred semi-insulating GaAs or InP. In x Ga 1x Sb was chosen for the base because of its small bandgap and to exploit the good hole transport characteris- tics of these alloys. In addition, low-resistance ohmic con- tacts to p-InGaSb have been achieved [150–152]. These properties are important factors in minimizing the base resistance. Equally important is the fact that a narrow bandgap InGaSb base can be used with a collector and an emitter made from a variety of InAlAsSb alloys. While little is known about the details of the band oﬀsets and bandgaps of these materials, extrapolations from known binary and ternary alloys indicate that it should be possible to have a large valence band oﬀset of 300 mV or more over a wide range of InAlAsSb alloys, particularly with alloys having a 6.2 A ˚lattice constant [3]. This is important in minimizing the parasitic hole current ﬂow from the InGaSb base to the emitter. The InAlAsSb alloys are predicted to be direct bandgap semiconductors over a range of bandgaps from 0.2 to 1.4 eV. This leads to a wide range of possible conduc- tion band oﬀsets with the InGaSb base because the valence band oﬀset is almost insensitive to the InAlAsSb composi- tion. The above listed properties allow a great deal of lati- tude in using band-structure engineering to optimize HBT performance. One possible emitter–base–collector structure with estimated band oﬀsets is illustrated in Fig. 22. The neg- ative side to using InAlAsSb alloys with the desired compo- sition is that they are expected to be hard to grow as controlling the As/Sb composition is diﬃcult, and there may be miscibility gap problems. Methods for obtaining n-type doping also are required. The structure in Fig. 22 was grown on an undoped GaSb substrate with a 1 lm AlSb buﬀer layer [153]. The collector was 1 lm thick with a nominal n=3·10 16 /cm 3 doping; the base was 0.1 lm thick with a nominal p=5·10 18 /cm 3 doping; the emitter was 0.2 lm thick with a nominal n=3·10 17 /cm 3 doping. The heterostructure also included a heavily doped 0.1 lm layer grown between the collector and the AlSb buﬀer. This layer has the same composition as the collector and it was included to improve the electrical contact to the collector. Heavily doped contact layers have been used on the top of the emitter to aid in contacting it. Rather conservative layer designs were used in this struc- ture. A relatively low base doping was chosen to avoid pos- sible problems associated with Be segregating into the emitter. A thick base was used to make it easier to etch down to it using an iterative etch test procedure. A thick collector was used to accommodate the lattice mismatch between the AlSb with a 6.136 A ˚lattice constant and the 6.2 A ˚HBT structure. A variety of emitter shapes were fab- ricated with relatively large areas of 30–60 lm 2 . The common emitter curves for the layer structure in Fig. 22 are shown in Fig. 23. It has a DC current gain of 25 and has been measured to V CE = 5 V without damage to the device. The maximum collector current, I C ,in Fig. 23 corresponds to a density of 1.8 ·10 4 A/cm 2 . The low collector–emitter oﬀset voltage of 220 mV supports the possibility of low power dissipation. Current–voltage measurements on the emitter–base and collector-base p-n junctions indicate that series resistances are spreading the voltage over which I C increases from zero in Fig. 23. Much higher collector currents densities are expected at lower base-emitter voltages for the InGaSb devices compared to Fig. 22. Predicted band diagram for InAlAsSb/InGaSb/InAlAsSb npn HBT with a o = 6.20 A ˚[149]. Fig. 23. IVcharacteristics from InAlAsSb/InGaSb/InAlAsSb npn HBT [153].IEE. 1890 B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 conventional InP or SiGe HBTs. This is important because large collector currents are required for high-frequency operation, and small values of V BE indicate that low power dissipation is possible. 4.4. pnp Structures and InP HBTs with GaAsSb bases With the exception of the 1992 paper by Pekarik et al., all the above work involves npn structures. Recently, Rockwell demonstrated a pnp structure with an InAs base and AlGaSb emitter and collector [154]. The bandstructure is shown in Fig. 24. The device exhibits gain for base-emit- ter voltages as low as 0.2 V. Our discussion here has focused on HBTs with very nar- row bandgaps and large lattice constants (6.0–6.2 A ˚), regardless of whether the heterostructures included Sb. There has been an important advance involving the use of Sb in InP-based HBTs. Speciﬁcally, the In 0.53 Ga 0.47 As base can be replaced with GaAs 0.51 Sb 0.49 to give a type-II band alignment to reduce collector-current blocking. Excellent HBT performance has been achieved with these structures [155–157]. 5. Summary and outlook In this paper, we have reviewed the recent progress on three antimonide-based electronic devices: HEMTs, R(I)TDs, and HBTs. For all three devices, the advantage of Sb-based structures over Si-, GaAs-, or InP-based struc- tures is the attainment of high-frequency operation with much lower power consumption. For the HEMT, recent advances include the demonstration of Ka- and W-band LNA circuits that operate at less than one-third the power of similar InP-based circuits. Applications may also include low-power logic; advantages and obstacles are discussed in a recent article [158]. Active areas of HEMT research include reduction of the gate-to-channel spacing for sub- 100-nm gate length and reduction of gate leakage currents. The R(I)TDs have been heavily studied and considered for application in MOBILE circuits, but, like their InP- and GaAs-based counterparts, are waiting for a viable commer- cial application. Spintronic concepts using RITDs both with and without ferromagnetic semiconductor materials have also been reported [159–163]. Several groups have been investigating HBTs based upon narrow-bandgap materials in the last three years. Most approaches involve lattice constants of 6.0–6.1 A ˚with InAs or In 0.86 Ga 0.14 As as the base and InAs or related alloys such as InAsP or InAlAs as the emitter and collector. Divider circuits have been reported. An alternative approach is to work at a 6.2–6.3 A ˚lattice constant using InGaSb as the base and InAlAsSb alloys as the emitter and collector. Future systems could also include unconventional devices that take advantage of the properties of antimo- nides. A good example of this is the backward diode being developed for use in THz imaging [133]. THz source con- cepts using these materials have also been suggested and need to be examined [164]. The large conduction band oﬀ- sets between AlSb and InAs and the large electron mobility in InAs make this system useful in the fabrication of one- dimensional quantum wires. Advances in the lateral pat- terning by C.H. Yang and colleagues have allowed them to demonstrate low-temperature electron elastic mean free paths greater than 1.4 lm and coherence lengths greater than 3 lm[165]. Such structures could form the basis of future electronic devices. Antimonide-based electronics may also beneﬁt from growth and processing advances resulting from work on electro-optic devices. The wide variety of bandgaps and oﬀ- sets along with the high mobilities are being exploited to develop long-wavelength (8–25 lm) superlattice detectors that are predicted to have superior performance to the HgCdTe detectors [166]. Lasers emitting in the 2–5 lm range are also being developed [167]. As we have discussed, there has been considerable pro- gress in Sb-based electronic devices in the last few years. The advances are especially signiﬁcant considering the rel- atively low level of funding and eﬀort, compared, for exam- ple, to InP-based electronics in the 1980s. In our view, it is unrealistic to think that Sb-based electronics will replace a large fraction of GaAs- or InP-based devices. However, they are promising for speciﬁc applications where both high-frequency operation and low power consumption are needed. System applications that have already transi- tioned from GaAs- to InP-electronics to achieve lower power consumption may be candidates for transition to antimonides. Acknowledgements The Oﬃce of Naval Research supported the develop- ment of this review as well as much of the NRL research discussed. The Defense Advanced Research Projects Agency (ABCS Program) also supported portions of the research. The authors thank colleagues at NRL including: Fig. 24. Band diagram for pnp HBT demonstrated by Rockwell [154]. IEEE. B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 1891 R. Bass, A.S. Bracker, P.M. Campbell, J.C. Culbertson, E.R. Glaser, M. Goldenberg, K.D. Hobart, D. McMor- row, N. Papanicolaou, D. Park, B.V. Shanabrook, B.P. Tinkham, R.J. Wagner, B.D. Weaver, and M.J. Yang; and collaborators at Northrop Grumman Corporation: W.R. Deal, A. Gutierrez, M.D. Lange, and R. Tsai. References [1] Nguyen LD, Larson LE, Mishra UK. Ultra-high-speed modulation- doped ﬁeld-eﬀect transistors—a tutorial review. Proc IEEE 1992; 80(4):494–518. [2] Dobrovolskis Z, Grigoras K, Krotkus A. Measurement of the hot- electron conductivity in semiconductors using ultrafast electric pulses. Appl Phys A-Mater 1989;48(3):245–9. [3] Vurgaftman I, Meyer JR, Ram-Mohan LR. Band parameters for III–V compound semiconductors and their alloys. J Appl Phys 2001; 89(11):5815–75. [4] Milnes AG, Polyakov AY. Indium arsenide—a semiconductor for high-speed and electrooptical devices. Mater Sci Eng B-Solid 1993; 18(3):237–59. [5] Milnes AG, Polyakov AY. Gallium antimonide device related properties. Solid-State Electron 1993;36(6):803–18. [6] Dutta PS, Bhat HL, Kumar V. The physics and technology of gallium antimonide: an emerging optoelectronic material. J Appl Phys 1997;81(9):5821–70. [7] Chang CA, Ludeke R, Chang LL, Esaki L. Molecular-beam epitaxy (MBE) of InGaAs and GaSbAs. Appl Phys Lett 1977;31(11):759–61. [8] Yano M, Suzuki Y, Ishii T, Matsushima Y, Kimata M. Molecular- beam epitaxy of GaSb and GaSbAs. Jpn J Appl Phys 1978;17(12): 2091–6. [9] Ludeke R. Electronic properties of (10 0) surfaces of GaSb and InAs and their alloys with GaAs. IBM J Res Dev 1978;22(3):304–14. [10] Bennett BR, Shanabrook BV. Molecular beam epitaxy of Sb- based semiconductors. In: Liu AWK, Santos MB, editors. Thin ﬁlms: heteroepitaxial systems. Singapore: World Scientiﬁc; 1999. p. 401–52. [11] Tuttle G, Kroemer H, English JH. Electron concentrations and mobilities in AlSb/InAs/AlSb quantum wells. J Appl Phys 1989; 65(12):5239–42. [12] Tuttle G, Kroemer H, English JH. Eﬀects of interface layer sequencing on the transport-properties of InAs/AlSb quantum- wells—evidence for antisite donors at the InAs/AlSb interface. J Appl Phys 1990;67(6):3032–7. [13] Bolognesi CR, Kroemer H, English JH. Well width dependence of electron-transport in molecular-beam epitaxially grown InAs/AlSb quantum-wells. J Vac Sci Technol B 1992;10(2):877–9. [14] Koester SJ, Bolognesi CR, Thomas M, Hu EL, Kroemer H, Rooks MJ. Determination of one-dimensional subband spacings in InAs/ AlSb ballistic constrictions using magnetic-ﬁeld measurements. Phys Rev B 1994;50(8):5710–2. [15] Yang CH, Yang MJ, Cheng KA, Culbertson JC. Characterization of one-dimensional quantum channels in InAs/AlSb. Phys Rev B 2002;66(11):115306. [16] Kuze N, Nagase K, Muramatsu S, Miya S, Iwabuchi T, Ichii A, et al. InAs deep quantum-well structures and their application to hall elements. J Cryst Growth 1995;150(1–4):1307–12. [17] Behet M, Nemeth S, De Boeck J, Borghs G, Tummler J, Woitok J, et al. Molecular beam epitaxy and characterization of InAs/AlGaSb heterostructures for magnetic sensing applications. Semicond Sci Technol 1998;13(4):428–32. [18] Johnson M, Bennett BR, Yang MJ, Miller MM, Shanabrook BV. Hybrid hall eﬀect device. Appl Phys Lett 1997;71(7):974–6. [19] Johnson M, Bennett BR, Hammar PR, Miller MM. Magnetoelec- tronic latching Boolean gate. Solid-State Electron 2000;44(6): 1099–104. [20] Nguyen C, Brar B, Kroemer H. Surface-layer modulation of electron concentrations in InAs–AlSb quantum-wells. J Vac Sci Technol B 1993;11(4):1706–9. [21] Boos JB, Kruppa W, Bennett BR, Park D, Kirchoefer SW, Bass R, et al. AlSb/InAs HEMTÕs for low-voltage, high-speed applications. IEEE Trans Electron Dev 1998;45(9):1869–75. [22] Venkatasubramanian R, Dorsey DL, Mahalingam K. Heuristic rules for group IV dopant site selection in III–V compounds. J Cryst Growth 1997;175:224–8. [23] Sasa S, Yamamoto Y, Izumiya S, Yano M, Iwai Y, Inoue M. Increased electron concentration in InAs/AlGaSb heterostructures using a Si planar doped ultrathin InAs quantum well. Jpn J Appl Phys 1 1997;36(3B):1869–71. [24] Zhao Y, Jurkovic MJ, Wang WI. Kink-free characteristics of AlSb/ InAs high electron mobility transistors with planar Si doping beneath the channel. IEEE Trans Electron Dev 1998;45(1):341–2. [25] Bolognesi CR, Dvorak MW, Chow DH. High-transconductance delta-doped InAs/AlSb HFETÕs with ultrathin silicon-doped InAs quantum well donor layer. IEEE Electron Dev Lett 1998;19(3):83–5. [26] Bennett BR, Yang MJ, Shanabrook BV, Boos JB, Park D. Modulation doping of InAs/AlSb quantum wells using remote InAs donor layers. Appl Phys Lett 1998;72(10):1193–5. [27] Bolognesi CR, Bryce JE, Chow DH. InAs channel heterostructure- ﬁeld eﬀect transistors with InAs/AISb short-period superlattice barriers. Appl Phys Lett 1996;69(23):3531–3. [28] Furukawa A, Mizuta M. Heterojunction bipolar-transistor utilizing AlGaSb/GaSb alloy system. Electron Lett 1988;24(22):1378–80. [29] Wood CEC. Surface exchange doping of MBE GaAs from S and Se captive sources. Appl Phys Lett 1978;33(8):770–2. [30] Subbanna S, Tuttle G, Kroemer H. N-type doping of gallium antimonide and aluminum antimonide grown by molecular-beam epitaxy using lead-telluride as a tellurium dopant source. J Electron Mater 1988;17(4):297–303. [31] Turner GW, Eglash SJ, Strauss AJ. Molecular-beam epitaxial- growth of high-mobility n-GaSb. J Vac Sci Technol B 1993;11(3): 864–7. [32] Bennett BR, Magno R, Papanicolaou N. Controlled n-type doping of antimonides and arsenides using GaTe. J Cryst Growth 2003; 251(1–4):532–7. [33] Bergman J, Nagy G, Sullivan G, Brar B, Kadow C, Lin H-K, et al. InAs/AlSb HFETs with f T and f max above 150 GHz for low-power MMICs. In: Proc InP related mater conf, 2003. p. 219–22. [34] Nguyen C, Ensslin K, Kroemer H. Magnetotransport in InAs/AlSb quantum-wells with large electron-concentration modulation. Surf Sci 1992;267(1–3):549–52. [35] Bennett BR, Tinkham BP, Boos JB, Lange MD, Tsai R. Materials growth for InAs high electron mobility transistors and circuits. J Vac Sci Technol B 2004;22(2):688–94. [36] Bolognesi CR, Caine EJ, Kroemer H. Improved charge control and frequency performance in InAs/AlSb-based heterostructure ﬁeld- eﬀect transistors. IEEE Electron Dev Lett 1994;15(1):16–8. [37] Tsai R, Barsky M, Boos JB, Bennett BR, Lee J, Papanicolaou NA, et al. Metamorphic AlSb/InAs HEMT for low-power, high-speed electronics. In: Proc IEEE GaAs IC symp, 2003. p. 294–7. [38] Miya S, Muramatsu S, Kuze N, Nagase K, Iwabuchi T, Ichii A, et al. AlGaAsSb buﬀer barrier on GaAs substrate for InAs channel devices with high electron mobility and practical reliability. J Electron Mater 1996;25(3):415–20. [39] Royter Y, Elliott KR, Deelman PW, Rajavel RD, Chow DH, Milosavljevic I, et al. High frequency InAs-channel HEMTs for low power ICs. Tech Digest IEDM 2003:30.7.1–4. [40] Boos JB, Bennett BR, Kruppa W, Park D, Mittereder J, Bass R, et al. Ohmic contacts in AlSb InAs high electron mobility transis- tors for low-voltage operation. J Vac Sci Technol B 1999;17(3): 1022–7. [41] Tsai R, Boos JB, Bennett BR, Lange MD, Grundbacker R, Namba C, et al. 275 GHz f max , 220 GHz f T AlSb/InAs HEMT technology. In: Dev res conf, 2004. p. 2: 12–3. 1892 B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 [42] Tsai R, Grundbacker R, Lange MD, Boos JB, Bennett BR, Nam P, et al. Manufacturable AlSb/InAs HEMT technology for ultra-low power millimeter-wave integrated circuits. In: GaAs mantech conf, 2004. p. 4. [43] Bergman J, Nagy G, Sullivan G, Ikhlassi A, Brar B, Kadow C, et al. Low-voltage, high-performance InAs/AlSb HEMTs with power gain above 100 GHz at 100 mV drain bias. In: Dev res conf, 2004. p. 243–4. [44] Yang MJ, Wang FC, Yang CH, Bennett BR, Do TQ. A composite quantum well ﬁeld-eﬀect transistor. Appl Phys Lett 1996;69(1):85–7. [45] Boos JB, Yang MJ, Bennett BR, Park D, Kruppa W, Yang CH, et al. 0.1 lm AlSb/InAs HEMTs with InAs subchannel. Electron Lett 1998;34(15):1525–6. [46] Lin HK, Kadow C, Dahlstrom M, Bae JU, Rodwell MJW, Gossard AC, et al. InAs/InAsP composite channels for antimonide-based ﬁeld-eﬀect transistors. Appl Phys Lett 2004;84(3):437–9. [47] Lin HK, Kadow C, Bae JU, Rodwell MJW, Gossard AC, Brar B, et al. Design and characteristics of strained InAs/InAlAs composite- channel heterostructure ﬁeld-eﬀect transistors. J Appl Phys 2005; 97(2):024505. [48] Yang MJ, Bennett BR, Fatemi M, Lin-Chung PJ, Moore WJ, Yang CH. Photoluminescence of InAsSb/AlSb single quantum wells: transition from type-II to type-I band alignment. J Appl Phys 2000; 87(11):8192–4. [49] Boos JB, Yang MJ, Bennett BR, Park D, Kruppa W, Bass R. Low- voltage, high-speed AlSb InAsSb HEMTs. Electron Lett 1999; 35(10):847–8. [50] Papanicolaou N, Tinkham BP, Boos JB, Bennett BR, Bass R, Park D. 6.2 A ˚In 0.2 Al 0.8 Sb/InAs 0.7 Sb 0.3 HEMTs for low-voltage high- frequency applications. In: Intl semicon dev res symp, 2003. [51] Tinkham BP, Bennett BR, Magno R, Shanabrook BV, Boos JB. Growth of InAsSb-channel high electron mobility transistor struc- tures. J Vac Sci Technol B 2005;23(4):1441–4. [52] Brar B, Kroemer H. Inﬂuence of impact ionization on the drain conductance in InAs–AlSb quantum-well heterostructure ﬁeld-eﬀect transistors. IEEE Electron Dev Lett 1995;16(12):548–50. [53] Longenbach KF, Beresford R, Wang WI. Application of split-gate and dual-gate ﬁeld-eﬀect transistor designs to InAs ﬁeld-eﬀect transistors. Solid-State Electron 1990;33(9):1211–3. [54] Boos JB, Kruppa W, Park D. Reduction of gate current in AlSb/ InAs HEMTs using a dual-gate design. Electron Lett 1996;32(17): 1624–5. [55] Bolognesi CR, Chow DH. InAs/AlSb dual-gate HFETÕs. IEEE Electron Dev Lett 1996;17(11):534–6. [56] Zhao Y, Jurkovic MJ, Wang WI. Enhancement-mode InAs n- channel high electron mobility transistors using Beryllium sheet doping. Solid-State Electron 1998;42(1):57–61. [57] Kadow C, Lin HK, Dahlstrom M, Rodwell M, Gossard AC, Brar B, et al. Reduction of the unintentional background electron density in AlSb/InAs/AlSb quantum wells. J Cryst Growth 2003;251(1–4): 543–6. [58] Triplett GE, Brown AS, May GS. Charge modiﬁcation in InAs/ AlGaSb HEMT structures. J Cryst Growth 2004;265(1–2):47–52. [59] Kadow C, Dahlstrom M, Bae JU, Lin HK, Gossard AC, Rodwell MJW, et al. n(+)-InAs–InAlAs recess gate technology for InAs- channel millimeter-wave HFETs. IEEE Trans Electron Dev 2005;52(2):151–8. [60] Bennett BR, Boos JB, Ancona MG, Papanicolaou N, Park D, Bass R. InAs quantum well heterostructures for FETs with ultra-short gates. In: Electron mater conf, 2005. [61] Kruppa W, Boos JB, Park D, Bennett BR, Bass R. Microwave noise characteristics of AlSb/InAs HEMTs. Electron Lett 1997;33(12): 1092–3. [62] Bergman J, Nagy G, Sullivan G, Brar B, Kadow C, Lin H-K, et al. RF noise performance of low power InAs/AlSb HFETs. In: Proc dev res conf, 2003. p. 147–8. [63] Deal WR, Tsai R, Lange MD, Boos JB, Bennett BR, Gutierrez A. A W-band InAs/AlSb low-noise/low-power ampliﬁer. IEEE Microw Wirel Co 2005;15(4):208–10. [64] Kruppa W, Boos JB, Bennett BR, Yang MJ. Low-frequency noise in AlSb/InAs HEMTs. Electron Lett 2000;36(22):1888–9. [65] Kruppa W, Yang MJ, Bennett BR, Boos JB. Low-frequency noise in AlSb/InAs high-electron-mobility transistor structure as a func- tion of temperature and illumination. Appl Phys Lett 2004;85(5): 774–6. [66] Kruppa W, Boos JB, Bennett BR, Tinkham BP. Low-frequency noise characteristics of AlSb/InAsSb HEMTs. Solid-State Electron 2004;48(10–11):2079–84. [67] Spitzer J, Hopner A, Kuball M, Cardona M, Jenichen B, Neuroth H, et al. Inﬂuence of the interlace composition of InAs/AlSb superlattices on their optical and structural-properties. J Appl Phys 1995;77(2):811–20. [68] Kramer GD, Adam MS, Tsui RK, Theodore ND. Characterization of dislocation reduction in MBE-grown (Al, Ga)Sb/GaAs by TEM. In: Gallium arsenide and related compounds, 1994. p. 727–32. [69] McMorrow D, Boos JB, Knudson AR, Buchner S, Yang MJ, Bennett BR, et al. Charge-collection characteristics of low-power ultrahigh speed, metamorphic AlSb/InAs high-electron mobility transistors (HEMTs). IEEE Trans Nucl Sci 2000;47(6):2662–8. [70] Weaver BD, Jackson EM. Universal behavior in irradiated high- electron-mobility transistors. Appl Phys Lett 2002;80(15):2791–3. [71] Weaver BD, Boos JB, Papanicolau NA, Bennett BR, Park D, Bass R. High radiation tolerance of InAs/AlSb high electron mobility transistors. Appl Phys Lett 2005;87:173501. [72] Hacker JB, Bergman J, Nagy G, Sullivan G, Kadow C, Lin HK, et al. An ultra-low power InAs/AlSb HEMT Ka-band low-noise ampliﬁer. IEEE Microw Wirel Co 2004;14(4):156–8. [73] Ashley T, Dean AB, Elliott CT, Pryce GJ, Johnson AD, Willis H. Uncooled high-speed InSb ﬁeld-eﬀect transistors. Appl Phys Lett 1995;66(4):481–3. [74] Goldammer KJ, Chung SJ, Liu WK, Santos MB, Hicks JL, Raymond S, et al. High-mobility electron systems in remotely- doped InSb quantum wells. J Cryst Growth 1999;202:753–6. [75] Ashley T. Novel InSb-based quantum well transistors for ultra-high speed, low power logic applications. In: Intl conf sol state & IC technol, 2004. [76] Sun JP, Haddad GI, Mazumder P, Schulman JN. Resonant tunneling diodes: models and properties. Proc IEEE 1998;86(4): 641–61. [77] Luo LF, Beresford R, Wang WI. Resonant tunneling in AlSb/InAs/ AlSb double-barrier heterostructures. Appl Phys Lett 1988;53(23): 2320–2. [78] Soderstrom JR, Chow DH, Mcgill TC. New negative diﬀerential resistance device based on resonant interband tunneling. Appl Phys Lett 1989;55(11):1094–6. [79] Schulman JN. Analysis of Sb-based resonant interband tunnel diodes for circuit modeling. Solid-State Electron 1999;43(8):1367–71. [80] Sugiyama H, Matsuzaki H, Oda Y, Yokoyama H, Enoki T, Kobayashi T. InP-based strained InGaAs/AlAs resonant tunneling diodes with high peak-current density and large peak-to-valley current ratio grown by metal-organic vapor-phase epitaxy. In: Proc InP and related mater conf, 2003. p. 211–4. [81] Soderstrom JR, Chow DH, Mcgill TC. Demonstration of large peak-to-valley current ratios in InAs/AlGaSb/InAs single-barrier heterostructures. Appl Phys Lett 1989;55(13):1348–50. [82] Soderstrom JR, Yao JY, Andersson TG. Observation of resonant tunneling in InSb/AlInSb double-barrier structures. Appl Phys Lett 1991;58(7):708–10. [83] Beresford R, Luo LF, Wang WI. Narrow-Gap InAs for hetero- structure tunneling. Semicond Sci Technol 1990;5:S195–9. [84] Collins DA, Yu ET, Rajakarunanayake Y, Soderstrom JR, Ting DZY, Chow DH, et al. Experimental-observation of negative diﬀerential resistance from an inas/gasb interface. Appl Phys Lett 1990;57(7):683–5. [85] Luo LF, Beresford R, Longenbach KF, Wang WI. Resonant interband coupling in single-barrier heterostructures of InAs/GaSb/ InAs and GaSb/InAs/GaSb. J Appl Phys 1990;68(6):2854–7. B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 1893 [86] Yang L, Chen JF, Cho AY. A new GaSb/AlSb/GaSb/AlSb/InAs double-barrier interband tunneling diode and its tunneling mecha- nism. J Appl Phys 1990;68(6):2997–3000. [87] Ting DZY, Collins DA, Yu ET, Chow DH, Mcgill TC. Large peak current densities in novel resonant interband tunneling heterostruc- tures. Appl Phys Lett 1990;57(12):1257–9. [88] Longenbach KF, Luo LF, Wang WI. Resonant interband tunneling in InAs/GaSb/AlSb/InAs and GaSb/InAs/AlSb/GaSb heterostruc- tures. Appl Phys Lett 1990;57(15):1554–6. [89] Houng MP, Wang YH, Shen CL, Chen JF, Cho AY. Improvement of peak-to-valley ratio by the incorporation of the InAs layer into the GaSb/AlSb/GaSb/AlSb/InAs double barrier resonant interband tunneling structure. Appl Phys Lett 1992;60(6):713–5. [90] Chen JF, Cho AY. The eﬀects of GaSb/InAs broken gap on interband tunneling current of a GaSb/InAs/GaSb/AlSb/InAs tunneling structure. J Appl Phys 1992;71(9):4432–5. [91] Chen JF, Cho AY. Interband tunneling between valence-band and conduction-band quantum-wells in a GaSb/AlSb/InAs/AlSb/GaSb/ AlSb/InAs triple-barrier structure. J Appl Phys 1992;72(3):960–3. [92] Brown ER, Parker CD. Resonant tunnel diodes as submillimetre- wave sources. Philos Trans Roy Soc A 1996;354(1717):2365–81. [93] Diamond SK, Ozbay E, Rodwell MJW, Bloom DM, Pao YC, Harris JS. Resonant tunneling diodes for switching applications. Appl Phys Lett 1989;54(2):153–5. [94] Goldman VJ, Tsui DC, Cunningham JE. Evidence for LO-phonon- emission-assisted tunneling in double-barrier heterostructures. Phys Rev B 1987;36(14):7635–7. [95] Li J, Mirabedini A, Mawst LJ, Savage DE, Matyi RJ, Kuech TF. Eﬀect of interface roughness on performance of AlGaAs/InGaAs/ GaAs resonant tunneling diodes. J Cryst Growth 1998;195(1–4): 617–23. [96] Dellow MW, Beton PH, Langerak CJGM, Foster TJ, Main PC, Eaves L, et al. Resonant tunneling through the bound-states of a single donor atom in a quantum-well. Phys Rev Lett 1992;68(11): 1754–7. [97] Kitabayashi H, Waho T, Yamamoto M. Resonant interband tunneling current in InAs/AlSb/GaSb/AlSb/InAs double barrier diodes. J Appl Phys 1998;84(3):1460–6. [98] Shen J. Mechanisms of valley currents in InAs/AlSb/GaSb resonant interband tunneling diodes. J Appl Phys 1995;78(10):6220–3. [99] Xu YY, Fay P, Chow DH, Schulman JN. Experimental investiga- tion of the temperature dependence of InAs–AlSb–GaSb resonant interband tunnel diodes. IEEE Trans Electron Dev 2004;51(7): 1060–4. [100] Tehrani S, SHen J, Goronkin H, Kramer G, Hoogstra M, Zhu TX. AlAs hole barriers in InAs/GaSb/AlSb interband tunnel diodes. In: Intl symp GaAs and related compounds. Inst Phys Conf Ser, vol. 136, 1993. p. 209–14. [101] Shiralagi K, Shen J, Tsui R. Eﬀects of layer design on the performance of InAs/AlSb/GaSb resonant interband tunneling diodes on GaAs substrates. J Electron Mater 1997;26(12):1417–21. [102] Magno R, Bracker AS, Bennett BR, Twigg ME, Weaver BD. The eﬀect of defects on InAs/AlSb/GaSb resonant tunneling diodes. In: Proc InP and related mater conf, 2000. p. 122–5. [103] Magno R, Weaver BD, Bracker AS, Bennett BR. Proton irradiation of InAs/AlSb/GaSb resonant interband tunneling diodes. Appl Phys Lett 2001;78(17):2581–3. [104] Schulman JN, Chow DH, Hasenberg TC. InAs/antimonide-based resonant-tunneling structures with ternary alloy layers. Solid-State Electron 1994;37(4–6):981–5. [105] Magno R, Bracker AS, Bennett BR. Resonant interband tunnel diodes with AlGaSb barriers. J Appl Phys 2001;89(10):5791–3. [106] Ozbay E, Bloom DM, Chow DH, Schulman JN. 1.7-Ps, Microwave, integrated-circuit-compatible InAs/AlSb resonant-tunneling diodes. IEEE Electron Dev Lett 1993;14(8):400–2. [107] Soderstrom JR, Brown ER, Parker CD, Mahoney LJ, Yao JY, Andersson TG, et al. Growth and characterization of high-current density, high-speed InAs/AlSb resonant tunneling diodes. Appl Phys Lett 1991;58(3):275–7. [108] Soderstrom JR, Chow DH, Mcgill TC. InAs/A1Sb double-barrier structure with large peak-to-valley current ratio—a candidate for high-frequency microwave devices. IEEE Electron Dev Lett 1990;11(1):27–9. [109] Smet JH, Broekaert TPE, Fonstad CG. Peak-to-valley current ratios as high as 50-1 at room-temperature in pseudomorphic InGaAs/ AlAs/InAs resonant tunneling diodes. J Appl Phys 1992;71(5): 2475–7. [110] Su YK, Chang JR, Lu YT, Lin CL, Wu KM, Wu ZX. Novel AlInAsSb/InGaAs double-barrier resonant tunneling diode with high peak-to-valley current ratio at room temperature. IEEE Electron Dev Lett 2000;21(4):146–8. [111] McMorrow D, Magno R, Bracker AS, Bennett BR, Buchner S, Melinger JS. Charge-collection dynamics of AlSb–InAs–GaSb resonant interband tunneling diodes (RITDs). IEEE Trans Nucl Sci 2001;48(6):1973–9. [112] Chow DH, Dunlap HL, Williamson W, Enquist S, Gilbert BK, Subramaniam S, et al. InAs/AlSb/GaSb resonant interband tunnel- ing diodes and Au-on-InAs/AlSb-superlattice Schottky diodes for logic circuits. IEEE Electron Dev Lett 1996;17(2):69–71. [113] Soderstrom JR, Chow DH, Mcgill TC. Observation of large peak- to-valley current ratios and large peak current densities in AlSb/ InAs/AlSb double-barrier tunnel structures. J Appl Phys 1989; 66(10):5106–8. [114] Chow DH, Schulman JN. Intrinsic current bistability in InAs/ AlGaSb resonant-tunneling devices. Appl Phys Lett 1994;64(1): 76–8. [115] Jimenez JL, Mendez EE, Li X, Wang WI. Resonant tunneling and intrinsic bistability in GaSb-based double barrier heterostructures. Solid-State Electron 1996;40(1–8):583–4. [116] Brown ER, Eglash SJ, Turner GW, Parker CD, Pantano JV, Calawa DR. Eﬀect of lattice-mismatched growth on InAs/AlSb resonant- tunneling diodes. IEEE Trans Electron Dev 1994;41(6):879–82. [117] Nosho BZ, Weinberg WH, Barvosa-Carter W, Bennett BR, Shana- brook BV, Whitman LJ. Eﬀects of surface reconstruction on III–V semiconductor interface formation: the role of III/V composition. Appl Phys Lett 1999;74(12):1704–46. [118] Lin Y, Gonzalez EM, Mendez EE, Magno R, Bennett BR, Bracker AS. Magnetotunneling between two-dimensional electron gases in InAs–AlSb–GaSb heterostructures. Phys Rev B 2003;68(3): 035311. [119] Nomoto K, Taira K, Suzuki T, Hase I. Experimental evidence of surface conduction in AlSb-InAs tunneling diodes. J Appl Phys 1999;85(2):953–8. [120] Nomoto K, Taira K, Suzuki T, Hase I, Hiroshima H, Komuro M. Diameter dependence of current–voltage characteristics of ultrasmall area AlSb-InAs resonant tunneling diodes with diameters down to 20 nm. Appl Phys Lett 1997;70(15):2025–7. [121] Reed MA, Randall JN, Aggarwal RJ, Matyi RJ, Moore TM, Wetsel AE. Observation of discrete electronic states in a zero-dimensional semiconductor nanostructure. Phys Rev Lett 1988;60(6):535–7. [122] Shiralagi K, Tsui R, Goronkin H, Pendharkar S, Tresek J, Allen S. Sidewall spacer deﬁned resonant interband tunneling diodes. J Electron Mater 1998;27(1):L1–4. [123] Williamson W, Enquist SB, Chow DH, Dunlap HL, Subramaniam S, Lei PM, et al. 12 GHz clocked operation of ultralow power interband resonant tunneling diode pipelined logic gates. IEEE J Solid-State Circ 1997;32(2):222–31. [124] Shen J, Kramer G, Tehrani S, Goronkin H, Tsui R. Static random- access memories based on resonant interband tunneling diodes in the InAs/GaSb/AlSb material system. IEEE Electron Dev Lett 1995; 16(5):178–80. [125] Shen J, Tehrani S, Goronkin H, Kramer G, Tsui R. An exclusive- nor based on resonant interband tunneling FETÕs. IEEE Electron Dev Lett 1996;17(3):94–6. 1894 B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 [126] Fay P, Lu J, Xu Y, Bernstein GH, Gonzalez A, Mazumder P, et al. Digital integrated circuit using integrated InAlAs/InGaAs/InP HEMTs and InAs/AlSb/GaSb RITDs. Electron Lett 2001;37(12): 758–9. [127] Bennett BR, Bracker AS, Magno R, Boos JB, Bass R, Park D. Monolithic integration of resonant interband tunneling diodes and high electron mobility transistors in the InAs/GaSb/AlSb material system. J Vac Sci Technol B 2000;18(3):1650–2. [128] Ancona MG, Boos JB, Justh EW. Modeling of ultra-low-power AlSb/InAs HEMT-RITD circuits. In: Proc InP and related mater conf, 2000. p. 130–3. [129] Hanbicki AT, Magno R, Cheng SF, Park YD, Bracker AS, Jonker BT. Nonvolatile reprogrammable logic elements using hybrid resonant tunneling diode-giant magnetoresistance circuits. Appl Phys Lett 2001;79(8):1190–2. [130] Brown ER, Soderstrom JR, Parker CD, Mahoney LJ, Molvar KM, Mcgill TC. Oscillations up to 712 Ghz in InAs/AlSb resonant- tunneling diodes. Appl Phys Lett 1991;58(20):2291–3. [131] Luo LF, Beresford R, Wang WI. Interband tunneling in poly- type GaSb/AlSb/InAs heterostructures. Appl Phys Lett 1989;55(19): 2023–5. [132] Chen JF, Yang L, Wu MC, Chu SNG, Cho AY. On the eﬀect of the barrier widths in the InAs/AlSb/GaSb single-barrier interband tunneling structures. J Appl Phys 1990;68(7):3451–5. [133] Schulman JN, Chow DH. Sb-heterostructure interband backward diodes. IEEE Electron Dev Lett 2000;21(7):353–5. [134] Fay P, Schulman JN, Thomas S, Chow DH, Boegeman YK, Holabird KS. High-performance antimonide-based heterostructure backward diodes for millimeter-wave detection. IEEE Electron Dev Lett 2002;23(10):585–7. [135] Meyers RG, Fay P, Schulman JN, Thomas S, Chow DH, Zinck J, et al. Bias and temperature dependence of Sb-based heterostructure millimeter-wave detectors with improved sensitivity. IEEE Electron Dev Lett 2004;25(1):4–6. [136] Vengurlekar AS, Capasso F, Chiu TH. Impact ionization in the base of a hot-electron AlSb/InAs bipolar-transistor. Appl Phys Lett 1990; 57(17):1772–4. [137] Pekarik JJ, Kroemer H, English JH. An AlSb–InAs–AlSb double- heterojunction P-N-P bipolar-transistor. J Vac Sci Technol B 1992; 10(2):1032–4. [138] Dodd PE, Lovejoy ML, Lundstrom MS, Melloch MR, Woodall JM, Pettit D. Demonstration of npn InAs bipolar transistors with inverted base doping. IEEE Electron Dev Lett 1996;17(4):166–8. [139] Moran PD, Chow D, Hunter A, Kuech TF. Fabrication of InAs/ AlSb/GaSb heterojunction bipolar transistors on Al 2 O 3 substrates by wafer bonding. Appl Phys Lett 2001;78(15):2232–4. [140] Averett KL, Maimon S, Wu X, Koch MW, Wicks GW. InAs-based bipolar transistors grown by molecular beam epitaxy. J Vac Sci Technol B 2002;20(3):1213–6. [141] Averett KL, Wu X, Koch MW, Wicks GW. Low-voltage InAsP/ InAs HBT and metamorphic InAs BJT devices grown by molecular beam epitaxy. J Cryst Growth 2003;251(1–4):852–7. [142] Maimon S, Averett KL, Wu X, Koch MW, Wicks GW. InAs-based heterojunction bipolar transistors. Electron Lett 2002;38(7):344–6. [143] Thomas SI, Elliott K, Chow DH, Shi B, Deelman P, Brewer P, et al. Fabrication and performance of InAs-based heterojunction bipolar transistors. In: Proc InP and related mater conf, 2003. p. 26–31. [144] Rieh JS, Jagannathan B, Greenberg DR, Meghelli M, Rylyakov A, Guarin F, et al. SiGe heterojunction bipolar transistors and circuits toward terahertz communication applications. IEEE Trans Microw Theory 2004;52(10):2390–408. [145] Kotani K, Yamabi R, Kawasaki T, Yanagisawa M, Yaegassi S, Yano H. High-speed and uniform self-aligned InGaAs/InP HBTs for 40 Gb/s ﬁber optic communications applications. Jpn J Appl Phys 1 2003;42(4B):2352–8. [146] Monier C, Sawdai D, Cavus A, Sandhu R, Lange M, Wang J, et al. High indium content metamorphic (In,Al)As/(In,Ga)As heterojunc- tion bipolar transistors. In: Proc InP and related mater conf, 2003. p. 32–5. [147] Noori AM, Sandhu RS, Hayashi SL, Meserole ED, Hardev V, Cavus A, et al. Strain relaxation and surface roughness of InAlAs graded buﬀer layers grown on InP for 6.05 angstrom applications. J Vac Sci Technol B 2004;22(5):2303–8. [148] Monier C, Cavus A, Sandhu RS, Lange MD, Cheng P, Chan B, et al. narrow-band-gap HBT technology for low-power, high-speed applications. In: GaAs mantech conf, 2004. p. 13.2. [149] Magno R, Bennett BR, Ikossi K, Ancona MG, Glaser ER, Papanicolaou N, et al. Antimony-based quaternary alloys for high-speed low-power electronic devices. Eastman conf, 2002. p. 288–96. [150] Wang SH, Mohney SE, Hull BA, Bennett BR. Design of a shallow thermally stable ohmic contact to p-type InGaSb. J Vac Sci Technol B 2003;21(2):633–40. [151] Wang SH, Mohney SE, Robinson JA, Bennett BR. Sulfur passiv- ation for shallow Pd/W/Au ohmic contacts to p-InGaSb. Appl Phys Lett 2004;85(16):3471–3. [152] Wang SH, Robinson JA, Mohney SE, Bennett BR. Shallow and thermally stable Pt/W/Au Ohmic contacts to p-type InGaSb. J Vac Sci Technol A 2005;23(2):293–7. [153] Magno R, Boos JB, Campbell PM, Bennett BR, Glaser ER, Tinkham BP, et al. InAlAsSb/InGaSb double heterojunction bipo- lar transistor. Electron Lett 2005;41(6):370–1. [154] Brar B, Bergman J, Pierson R, Rowell P, Nagy G, Sullivan G. Low- voltage AlGaSb/InAs/AlGaSb pnp HBTs. In: Dev res conf, 2003. p. 91–2. [155] Bolognesi CR, Matine N, Dvorak MW, Yeo P, Xu XG, Watkins SP. InP/GaAsSb/InP double HBTs: a new alternative for InP-based DHB. IEEE Trans Electron Dev 2001;48(11):2631–9. [156] Dvorak MW, Bolognesi CR, Pitts OJ, Watkins SP. 300 GHz InP/ GaAsSb/InP double HBTs with high current capability and BV CEO > = 6 V. IEEE Electron Dev Lett 2001;22(8):361–3. [157] Rajavel RD, Hussain T, Montes MC, Sawins MW, Thomas S, Chow DH. Molecular beam epitaxial growth and characterization of InP/GaAsSb/InP double heterojunction bipolar transistors. J Cryst Growth 2003;251(1–4):848–51. [158] Chau R, Datta S, Doczy M, Doyle B, Jin J, Kavalieros J, et al. Benchmarking nanotechnology for high-performance and low- power logic transistor applications. IEEE Trans Nanotechnol 2005;4(2):153–8. [159] Vurgaftman I, Meyer JR. Ferromagnetic resonant interband tun- neling diode. Appl Phys Lett 2003;82(14):2296–8. [160] Moon JS, Chow DH, Schulman JN, Deelman P, Zinck JJ, Ting DZY. Experimental demonstration of split side-gated resonant interband tunneling devices. Appl Phys Lett 2004;85(4):678–80. [161] Yamada S, Kikutani T, Gozu S, Sato Y, Kita T. Spontaneous spin- splitting observed in resonant tunneling diode with narrow band-gap asymmetric quantum well. Physica E 2002;13(2–4):815–8. [162] Koga T, Nitta J, Takayanagi H. Spin-ﬁlter device based on the Rashba eﬀect using a nonmagnetic resonant tunneling diode. Phys Rev Lett 2002;88(12):126601. [163] Petukhov AG, Demchenko DO, Chantis AN. Electron spin polar- ization in resonant interband tunneling devices. Phys Rev B 2003; 68(12):125332. [164] Woolard D, Zhang WD, Gelmont B. A novel interband-resonant tunneling diode (I-RTD) based high-frequency oscillator. Solid- State Electron 2005;49(2):257–66. [165] Yang CH, Yang MJ, Cheng KA, Culbertson JC. A novel InAs quantum wire system. Physica E 2003;17(1–4):161–3. [166] Fuchs F, Weimer U, Pletschen W, Schmitz J, Ahlswede E, Walther M, et al. High performance InAs/GaInSb superlattice infrared photodiodes. Appl Phys Lett 1997;71(22):3251–3. [167] Meyer JR, Hoﬀman CA, Bartoli FJ, Rammohan LR. Type-II quantum-well lasers for the midwavelength infrared. Appl Phys Lett 1995;67(6):757–9. B.R. Bennett et al. / Solid-State Electronics 49 (2005) 1875–1895 1895 ... InSb as a semiconductor (III-V semiconductors). [334][335][336][337][338] Another example is Cu2Sb which is used to test for the oxidation of antibiotic drugs. [339] Furthermore, it is also discussed as an anode material for Li-ion batteries, [340][341][342][343][344][345][346] as an anode material for Na-ion batteries [347][348][349] and in Na-ion fuel cells. ... Thesis Full-text available See the full Abstract and information on: https://nbn-resolving.org/urn:nbn:de:bsz:14-qucosa2-804979 ... Ternary antimonide films such as In 1-x Ga x Sb [1] and In 1-x Al x Sb [2] offer great advantages given that their bandgap energy and lattice parameters can be easily tuned, between those of their binary counterparts, by simply varying their stoichiometry. In the case of In 1-x Ga x Sb, the bandgap energy can attain values between 0.17 eV (InSb) and 0.85 eV (GaSb), at 300 K, with possible applications in high speed and low power consumption electronic devices [3]. In 1-x Ga x Sb also finds applications in thermoelectrics (TE) [4], which is particularly interesting given the urgent need for alternative ways to produce clean energy. ... Article This article reports on the atomic composition and structural changes induced by ion irradiation in In1-xGaxSb films deposited by magnetron sputtering on SiO2/Si. Samples with x values equal 0, 0.2, 0.4, 0.5 and 1 were irradiated with 16 MeV Au⁺⁷ ions, in the fluence range 5 × 10¹³-2 × 10¹⁴ cm⁻² (3 × 10¹⁴ for GaSb) and the structure and atomic composition of the films were investigated. Upon irradiation, all films attain a nanofoam-like structure, and the most pronounced swelling was observed in ternary films with 20% Ga atomic concentration. With particle induced x-ray emission technique, we identified the presence of C, O, Ga, In and Sb, with C and O concentrations significantly higher in the nanofoams, compared to the as-deposited films. Rutherford backscattering spectrometry analysis showed the atomic composition of the ternaries is not uniform, but forms two layers with slightly different relative atomic concentrations, specially after irradiation, with C and O uptake greatly enhanced by ion irradiation, more pronounced towards the surface. Grazing incidence x-ray diffraction analysis revealed that ion irradiation with total fluence of 2 × 10¹⁴ cm⁻² induces amorphization of the ternaries, except for samples with 50% Ga, which remain polycrystalline, despite the ion-induced porosity. Article Barrier infrared photodetectors have been proposed to improve the device performance, such as the operating temperature. In this article, a quantum tunneling barrier with alternating AlAs0.08Sb0.92 and AlSb layers is adopted to enhance carrier extraction from an InAs0.91Sb0.09 (InAsSb) absorber layer (AL). Due to the small valence band offset between the AlSb and InAsSb layers, the photogenerated holes can be efficiently extracted by tunneling, despite the hole barrier between AlAs0.08Sb0.92 and InAsSb layers. The applied voltage of the detector with a tunneling barrier structure is reduced by about 31%, compared with that of a reference detector with an AlAs0.08Sb0.92 barrier, to fully extract the photogenerated holes. At 150 K, the quantum tunneling barrier detector exhibits a dark current density of ~4.35\times 10^{{-{5}}}$A/cm2 and a quantum efficiency (QE) of 50.57% under 3.3-$\mu \text{m}$radiation corresponding to a responsivity of 1.34 A/W at a low bias voltage of −0.18 V. Due to the tunneling barrier, a peak specific detectivity as high as$3.60\times 10^{{11}}$cm$\cdot $Hz$^{\text {1/2}}\$ /W is achieved at a relatively low bias voltage of −0.14 V.
Article
The InAs/GaSb superlattices (SPLs) was an important component of Quantum Cascade Laser (QCL) and Interband Cascade Laser (ICL). In particular, the upper and lower SPLs waveguide layers and active regions of the ICL were alternately grown from a large number of ultra-film epitaxial layers (nm) by Molecular Beam Epitaxy(MBE). Subtle lattice mismatch may directly lead to the deterioration of material crystal quality, and the thickness, which the composition change of each layer will strongly affect the structural performance of device materials. The optimal growth temperature of InAs/GaSb SPLs were about 420℃. By growing 40× short period GaSb/AlSb and InAs/GaSb SPLs with the substrate rotating, the thickness of GaSb and AlSb layers were 5.448 nm and 3.921 nm, and the thickness of InAs and GaSb layers were 8.998 nm and 13.77 nm, respectively. The error was about 10%, and the optimal growth conditions of InAs/AlSb SPLs were obtained. Due to the lattice matched, the 40× InAs/away as injection on the average lattice constant of InAs/AlSb SPLs were fully considered. Under the condition that the soak time was fixed by 3s, the average lattice constant of SPLs was adjusted by changing the as pressure to 1.7e⁻⁶ mbar to achieve lattice matching on the GaSb substrate. The results showed that the 0 order satellite peak of the SPLs coincides with the peak of the GaSb substrate, indicated perfect lattice matching, and the sharp of second order satellite peak also showed excellent structural quality of the SPLs structure.
Article
Ionic liquids (ILs) are able to activate elements that are insoluble in common solvents. Here, the synthesis of binary antimony compounds directly from elements was explored. The 12 elements Ti–Cu, Al, Ga, In, and Te, known to form binary compounds with Sb, were reacted with Sb in [P 66614 ]Cl under inert conditions in a closed glass flask with vigorous stirring for 16 h at 200 °C. This was immediately successful in four cases and resulted in the formation of NiSb, InSb, Cu 2 Sb and Sb 2 Te 3 . The applied reaction temperature is several hundred degrees below the temperatures required for solvent‐free conversions. Compared to reactions based on diffusion in the solid state, reaction times are much shorter. The IL is not consumed and can be recycled. Since the reaction with Cu showed almost complete conversion, the influences of reaction time, temperature and medium were further investigated. Among the tested imidazolium ILs ([BMIm]Cl, [BMIm][OAc], [BDMIm]Cl) and phosphonium ILs ([P 66614 ] X , X = Cl – , [DCA] – , [OAc] – , [NTf 2 ] – ), those with chloride anion yielded the best results. In a diffusion experiment, Cu 2 Sb formed on the copper, which indicates that antimony forms mobile species in these ILs. Supplemental crystal structure data of (As 3 S 4 )[AlCl 4 ], which was ionothermally synthesized from As and S, are reported.
Article
InAs/InP0.69Sb0.31 superlattice (SL) has been successfully grown on InAs substrate using metalorganic chemical vapor deposition for short-wavelength infrared detection. The SL epilayer has a fairly good structural quality and surface morphology evidenced by X-ray diffraction and atomic force microscopy. A strong 2.83μm peak was observed in photoluminescence (PL) measurement at 77 K, deviated from the theoretical estimation. Scanning transmission electron microscope (STEM) reveals well-arranged sublayers and asymmetric interfaces. The incorporation of arsenic into the InPSb layers due to As-carryover effect and As/Sb exchange was proposed to explain the reduced effective bandgap and SL interfacial asymmetry.
Chapter
High Electron Mobility Transistor (HEMT) attained great interest because of its superior electron transport making it suitable for applications in high-speed circuits and high power requirements. These devices are finding special interest to replace conventional field-effect transistors having outstanding performance in the domain of high-frequency applications. In HEMT, the high mobility of electrons and highly confined characteristics of the two-dimensional electron gas made sure that modulation doping could be utilized to have high-speed field-effect transistors having brilliant “Short Channel Effects” (SCEs) and excessive scope of scaling. However, lack of existing experimental results of such a device, designers require a dependable tool for simulation and analysis of the device characteristics in less time and low cost before device is fabricated for commercial use. Therefore, physics-based device simulator for design and performance prediction of the semiconductor device are very important. This book chapter describes an overview of the HEMT device and its physics-based simulation for performance analysis.
Chapter
The vapor phase growth of metal oxides has been utilized in a wide range of areas including electronics, optics, and protective coating technologies. Within this short review, we introduce the basic concepts of vapor phase growth, in particular chemical vapor deposition (CVD) and atomic layer deposition (ALD), as scalable growth processes to achieve demanding technology requirements. We then focus this article toward the miniaturization of electronic devices, where vapor phase processes can enable high‐quality material growth, at the atomic layer of precision, both in terms of thickness and conformality over complex device topographies. To illustrate these points, we describe how the specific advantages of these deposition techniques have been applied to the growth of two material classes, namely, ferroelectric oxides and dielectric oxides and discuss where vapor phase growth has made, or has the potential to make, significant advances in technology.
Article
Full-text available
Ohmic contacts to p-type InGaSb have been investigated. The factors that influence the contact resistance, thermal stability, and shallowness of the contacts are examined. The most desirable contact studied in this work employs three layers. A very thin layer of palladium is deposited on the p-InGaSb first and is found to lower the resistance at the metal/semiconductor interface. The next layer is W, which is predicted to be in thermodynamic equilibrium with InGaSb and which serves as a diffusion barrier to protect the semiconductor from the reaction with the final capping layer. The final capping layer is a 100 or 150 nm Au layer. The Au lowers the metal sheet resistance, which we have found both experimentally and through modeling to influence the contact resistance measurements, and the Au layer provides a contact surface that does not oxidize. The contact resistance of the as-deposited Pd/W/Au (5/50/145 nm) contact is 0.08 Ω mm (corresponding to a specific contact resistance of ≪3×10-7 Ω cm2), while the more thermally stable Pd/W/Au (5/145/100 nm) contact exhibits a contact resistance of 0.08 Ω mm only after annealing at 250 °C for 3 h, in both cases on a p-In0.25Ga0.75Sb layer with a semiconductor sheet resistance of approximately 300 Ω/◻. The thermal stability of the Pd/W/Au contacts was also examined. The Pd/W/Au (5/145/100 nm) contacts remained shallow and exhibited no measurable electrical degradation when aged at 250 °C in N2 for 100 h, while they survived at 250 °C for 14 days in sealed, evacuated, quartz tubes. © 2003 American Vacuum Society.
Article
Full-text available
We have observed that the tunneling magnetoconductance between two-dimensional (2D) electron gases, formed at nominally identical InAs-AlSb interfaces, most often exhibits two sets of Shubnikov–de Haas oscillations with almost the same frequency. This result is explained quantitatively with a model of the conductance in which the 2D gases have different densities and can tunnel between Landau levels with different quantum indices. When the epitaxial growth conditions of the interfaces are optimized, the zero-bias magnetoconductance shows a single set of oscillations, thus proving that the asymmetry between the two electron gases can be eliminated.
Conference Paper
The peak-to-valley current (P/V) ratio in InAs/AlSb/GaS/AlSb/InAs resonant interband tunnel diodes was increased by inserting monolayers of AlAs adjacent to the AlSb barriers. The highest P/V ratio without the AlAs barriers was 15 while the highest P/V ratio with the AlAs barriers was 30 at room temperature. The AlAs layer acts as a barrier to the parasitic conduction of the holes that are confined in the GaSb quantum well.
Conference Paper
High dislocation densities result when InAs or (Al,Ga)Sb epitaxial layers are grown on GaAs substrates due to the large lattice mismatch (similar to 8%). Cross-sectional Transmission Electron Microscopy (TEM) has been used to study the distribution of these dislocations. For InAs and AlSb layers the dislocation density decreases quickly for the first 2 mu m, then slowly decreases to 3 x 10(8)/cm(2) after 10 mu m of epi growth. In a thin-element superlattice (GaSb/AlSb, 10 nm periods) the dislocation densities are comparable to those of the bulk InAs and AlSb layers. When the layer thickness of the same periodic structure is increased to 100 nm the dislocation density is reduced by an order of magnitude to 4 x 10(7)/cm(2) after 10 mu m of epi.
Article
Very low-power InAs/AlSb HFETs with excellent RF performance are reported. These metamorphic HFETs on GaAs substrates combine high microwave gm of at least 1.1 S/mm with low parasitic resistances to offer simultaneous measured fτ and fmax values of 160 GHz for both figures of merit. This performance is obtained at a drain bias voltage of only 0.35 V for an HFET with a 0.25-μm gate length. The high current gain (fτ) is attributable to the improved charge control due to scaling of the barrier thickness to 180 Å. The maximum power gain (fmax) depends on both gm and the HFET output conductance, which is fundamentally limited by the low breakdown voltage gap of the InAs channel (Eg = 0.36 eV).
Article
Enhancement-mode InAs n-channel high electron mobility transistors (E-HEMT's) are realized by incorporating a Beryllium (Be) doping sheet within the upper barrier and utilization of an InAs surface layer. At room temperature, n-channel E-HEMT's with 1 mu m gate length exhibit extremely low output conductance (12 mS mm(-1)), high extrinsic transconductance (425 mS mm(-1) for V-DS = 0.8 V), and near zero threshold voltage. Our results demonstrate enhancement-mode operation of an InAs/AlSb heterojunction field-effect transistor. (C) 1998 Elsevier Science Ltd.
Article
Large mobilities and electron saturation velocity make InAs a promising material for high speed devices. Investigations into materials characteristics of doped InAs show nonideal behavior with standard molecular beam epitaxy dopants, silicon, and beryllium. Critical thicknesses for cracking of AlxIn1-xAs on InAs were empirically determined as a function of x. Mesa pn junctions in InAs show no effects of surface Fermi level pinning and exhibit good rectification with low reverse leakage. Bipolar junction transistor and heterojunction bipolar transistor devices are presented, along with their dc electrical characteristics. Common emitter current gains of 100 have been achieved in these bipolar devices. © 2002 American Vacuum Society.
Article
A comprehensive study of the transport dynamics associated with electron- and hole-charge residing within a type-II AlGaSb/InAs/InGaSb double-barrier heterostructure is presented. These studies demonstrate an intrinsic instability within this type of resonant tunneling diode (RTD) that is initiated by an interband tunneling process. This novel interband-RTD (I-RTD), based upon staggered-bandgap heterostructures, is shown to offer a completely new avenue for the realization of very high-frequency sources. Specifically, this I-RTD device is shown to admit hole generation and discharging processes leading to repetitive cycles in electron current that constitute steady-state oscillatory behavior. Initial studies of non-optimized structures predict impressive figures of merit for oscillation frequencies (e.g., ∼300 GHz) and output powers (e.g., >2 mW) and suggest that the I-RTD is a promising new device for implementation as a terahertz (THz) frequency oscillator. Furthermore, these studies suggest that strategic engineering of the interband tunneling and charge accumulation processes may allow for significant latitude in controlling the performance of this oscillator concept at very high-frequencies.