RELATED APPLICATIONS This application claims priority to U.S. Provisional Application Serial No.

61/168,434, entitled "BIPOLAR TRANSISTOR WITH QUANTUM WELL BASE AND QUANTUM WELL EMITTER," filed April 10, 2009.

DISCUSSION OF RELATED ART A bipolar junction transistor typically has three terminals: a base, an emitter, and a collector. The main current of the transistor flows between the emitter and collector terminals. The base is a control terminal that controls the flow of current between the emitter and the collector.

The development of monolithic microwave integrated circuit technology has pushed Bipolar Junction Transistors (BJT) to their limits in terms of frequency and gain. The development of Heteroj unction Bipolar Transistors (HBT) allowed higher emitter efficiency due to the energy gap difference between the emitter and base, which are formed of different materials. Increased base doping may also allow these transistors to perform at a higher frequencies. Although HBTs are widely used in the microwave industry, there is a need to improve the gain and noise performance of these devices.

A quantum well is a potential well that can confine the wavefunction of carriers within a region. A quantum well can be formed as a thin layer of material having a potential different from the adjacent layers. Combining a thin base with various heterostructures can allow a designer to leverage quantum effects to achieve favorable properties. Some prior work has been performed in this area. Capasso and Kiehl proposed a

Resonant tunneling Bipolar Transistor (RBT) having a double barrier and a quantum well in the base to form a negative transconductance device. Continuing with the same concept, the Bipolar Quantum Resonant Tunneling Transistor (BiQuaRTT) was developed, which has a quantum well base and two barriers at both junctions. Narrow Base HBTs (NBHBT) have a reduced base thickness which lessens both the base transit time and the base recombination current, thus enhancing DC current gain. The Bipolar Inversion Channel Field Effect Transistor (BICFET) is an example of a NBHBT which has a β of 10 ⁵ and high current operation (10 ⁶ A/cm ² j. This device has no base to limit scaling in the vertical dimension as with a bipolar transistor, and it has no drain to limit its scaling in the planar dimension as with a Field Effect Transistor (FET). Another NBHBT proposed by K. Ikossi-Anastasiou et al. has a base thickness of just 50 A. This transistor exhibits a maximum small-signal common emitter current gain of 1400 at 300° K and 3000 at 80° K. The Heterostructure Emitter and Heterostructure Base Transistor (HEHBT) with pseudomorphic base has a quantum well base of Ino _.2 Gao _.8 As enhancing the valence band discontinuity (ΔEv), which results in a higher emitter injection efficiency. The device shows S-shaped negative differential resistance phenomena in the inverted operation mode. Quantum well bases are also used for optoelectronic purposes. M. Feng et al. reports enhanced radiative recombination realized by incorporating InGaAs quantum wells in the base layer of Light- Emitting InGaP/GaAs heteroj unction bipolar Transistors (LETs) operating in the common- emitter configuration. Such devices simultaneously show both an amplified electrical output and an optical output with signal modulation. A paper by A.C.Seabaugh (Quantum-well resonant tunneling transistors", Proceedings IEEE/Cornell Conference on Advanced Concepts in High Speed Semiconductor Devices and Circuits, p. 255 - 264, 1989) gives a summary of several kinds of resonant tunneling HBTs that have been developed.

SUMMARY

Some embodiment relate to a bipolar transistor. The bipolar transistor may include an emitter that includes a first semiconductor region of a first conductivity type. The first semiconductor region has a small enough thickness to form a first quantum well. The bipolar transistor may also include a collector that includes a second semiconductor region of the first conductivity type. The bipolar transistor may further include a base comprising a third semiconductor region of a second conductivity type. The base is formed between the first semiconductor region and the second semiconductor region. The third semiconductor region has small enough thickness to form a second quantum well.

In some aspects, the first quantum well is formed of a material of higher bandgap than the second quantum well.

In some aspects, the third semiconductor region is formed of a different semiconductor material than the first and second semiconductor regions.

In some aspects, the first and second semiconductor regions are formed of the same semiconductor material.

In some aspects, the first and second semiconductor regions comprise InGaP and the third semiconductor region comprises GaAs. In some aspects, the first and second semiconductor regions have a first conductivity type and the third semiconductor region has a second conductivity type opposite to the first conductivity type. In some aspects, due to the presence of the first and second quantum wells, the current gain of the transistor is better than in a corresponding HBT without a quantum well, wherein the HBT has an emitter, base and collector formed of the same materials as the bipolar transistor.

In some aspects, the first and second quantum wells significantly improve the breakdown voltage of the transistor.

BRIEF DESCRIPTION OF THE DRAWINGS

As is conventional in the representation of semiconductor devices, the drawings are not necessarily drawn to scale for purposes of illustration. FIG. 1 shows a cross section of a bipolar transistor according to one embodiment.

FIG. 2 shows a top-view image of the transistor of FIG. 1. FIG. 3 shows a graph of I- V characteristics for the transistor of FIG. 1. FIG. 4 shows a Gummel plot for the transistor of FIG. 1. FIG. 5 shows a cross section of a bipolar transistor with a quantum well base. FIG. 6 shows plots of the collector current versus collector voltage for various base current values of a commercially available hetrojunction bipolar transistor.

FIG. 7 shows plots of the collector current vs. collector voltage for various base current values of a quantum well base hetrojunction bipolar transistor.

FIG. 8 shows plots of the S ₂ i parameter for the quantum well base hetrojunction bipolar transistor.

FIG. 9 shows a Gummel plot comparison for the commercial HBT and QWBHBT. FIG. 10 shows a plot of the current gain vs. collector current for the commercial HBT. FIG. 11 shows a plot of current gain vs. collector current for the QWBHBT. FIG. 12 shows I-V curves for the commercial HBT at a temperature of 150 ⁰ K. FIG. 13 shows I-V curves for the QWHBT at a temperature of 15O ⁰ K.

DETAILED DESCRIPTION

The techniques described herein relate to transistors, such as bipolar junction transistors (BJTs), having both a quantum well base and a quantum well emitter. Such a transistor can achieve high current gain, better breakdown voltage and higher operational frequency. In addition, tight quantizing control of electron flow can reduce the noise level in the device. Prior attempts to manufacture quantum well base BJTs were focused on producing negative resistance devices. The transistor structure described herein may be considered to be a double heteroj unction bipolar transistor (DHBT) with a quantum well base. The energy band profile of a DHBT can be used to create a quantum well for holes and a quantum barrier for electrons. In the absence of a quantum barrier in the base of an HBT, electrons from the emitter can travel through the base towards the collector with different velocities and energies. The use of a quantum barrier can restrict or filter the allowable energy levels of the electrons passing through the barrier, resulting in a reduction of noise. Also, by maintaining sufficient spacing between the energy levels in the quantum barrier, scattering due to optical phonons can be reduced, further reducing noise. The reduction in base width can also result in a large gain for the transistor. An embodiment of a transistor 1 having a quantum well base and a quantum well emitter is shown in FIG. 1. FIG. 2 shows an image of the transistor 1 after fabrication. The base region 2 of the transistor 1 can be formed of a III- V semiconductor such as p-type doped GaAs having a doping concentration of 7*10 ¹⁸ cm ^'3 . As shown in FIG. 1, the thickness of this base region 2 may be about 240 A, such that the thickness is comparable to the wavelength of the electrons passing through it. The term "thickness" in this context refers to the effective base width which excludes the depletion region widths from the emitter-base junction and collector-base junction. This is only one example of a suitable thickness for the base region 2, as other thicknesses of the GaAs base region may be used that are the same as, substantially the same as, the wavelength of the electrons passing through this region. The GaAs base region 2 can form a quantum well that confines the electrons in this region. The combination of the lower energy gap p-type GaAs base region 2 sandwiched between two higher energy gap n-type InGaP regions 3, 4, can create a double heterojunction that results in the formation of a deep quantum well for the holes injected into the base from the base contact, and a quantum barrier for the electrons injected from the emitter through the base into the collector. An n-type doped InGaP region 3 of the emitter, adjacent to the base region 2, is also shown in FIG. 1. This InGaP region 3 of the emitter may have a thickness of 21θA, or other suitable thickness, forming a quantum well with respect to the electrons passing between the emitter and the collector. The base region 2 thus forms a first quantum well for the base, and the emitter InGaP region 3 forms a second quantum well for the emitter. These quantum wells may reduce the noise created by both electrons and holes. As shown in FIG. 1, the emitter may also have an n-type doped GaAs region 6 in contact with the InGaP region 3. The GaAs region 6 may have a thickness of about 2000A or other suitable thickness. The emitter may further include a highly n-type doped InGaAs region 7, which may have a thickness of about 650 A or other suitable thickness. An emitter electrode 8 may be in contact with the InGaAs region 7. A base electrode 9 may be in contact with the base region 2.

An additional InGaP region 4 in the collector adjacent to the base is also shown in FIG. 1. The InGaP region 4 of the collector may be 4OθA thick. In the embodiment of FIG. 1 , the collector has not been entirely formed of InGaP because it may be difficult to grow a sufficiently thick layer of InGaP in some circumstances. In the embodiment of FIG. 1, the remaining sub collector and collector regions may be formed of a GaAs region 5 having a doping concentration of 2*10 ¹⁶ cm ^'3 and 5*10 ¹⁸ cm ^"3 , respectively. However, other embodiments may use a collector layer entirely formed of InGaP formed using suitable manufacturing techniques. A collector electrode 10 may be formed in contact with GaAs region 5.

The transistor was modeled with SILVACO Technology Computer Aided Design (TCAD) software, which is used to model semiconductor fabrication and semiconductor device operation. Measurement of I- V (current- voltage) characteristics of the Quantum Emitter and Base HBT (QEBHBT) are shown in FIG. 3 and a Gummel plot is shown in FIG. 4. Gummel curves show that an increase in the built-in potential for the base emitter junction can be realized. The transistor can also have a higher collector offset voltage. The current gain of the transistor is around 350. The breakdown voltage of produced devices was also tested. The measurement of the breakdown voltage Vbr starts from bias point (V _c = 3 V and collector current I _c = 20 mA). The measurement is performed by increasing the collector voltage in small steps until the collector current shoots up quickly and the transistor fails catastrophically. The corresponding collector voltage at the point of breakdown was recorded as the breakdown voltage. Using these techniques, a breakdown voltage of 19V was measured for the transistor, which indicates an improvement by almost 58% as compared to the breakdown voltage of 12V for a commercial HBT.

Example 1

A heterostructure bipolar transistor (HBT) was modeled where the base is designed as a quantum well. The transistor structure includes a base layer of GaAs that is heavily p-type doped. The 0.024 μm base is sandwiched between wide band gap In _x Ga] _-X P which forms the n-type emitter and collector layers immediately adjacent to the base. The thickness of the base was chosen so that it is comparable to the wavelength of the electrons passing though it. There are two heavily doped cap layers of InGaAs at the emitter contact. The remainder of the emitter and collector regions are formed of GaAs. One goal of this design is to filter the energies and velocities of electrons as they pass through the base region that forms a quantum barrier to electrons and a quantum well to holes. This results in a significant decrease of noise in comparison to that observed in non-quantum base HBTs. The thin quantum well improves the collection of injected carriers, which in turn boosts the DC gain (β) to 750 and increases the power of the transistor by a factor of six, in comparison to a commercially available HBT with a similar non-quantum well structure. At high frequencies, the gain of the device is increased by about 5 dB over the non-quantum base HBT. Additionally, the cutoff frequency is improved from 20 GHz to 50 GHz. Modeling of the transistor was done using Silvaco ATLAS software. The dimensional characteristics of a bipolar device's base are a major factor in determining its performance characteristics. Combining a thin base with various heterostructures allows the designer to leverage quantum effects to achieve favorable properties.

In this example, we propose a structure that is essentially a double heterojunction bipolar transistor (DHBT) with a quantum well base. The concept is to use the existing energy band profile of a DHBT to create a quantum well with regards to the holes and a quantum barrier with regards to the electrons. In the absence of a quantum barrier in the base of an HBT, electrons from the emitter will travel through the base towards the collector with different velocities and energies. The presence of the quantum barrier will restrict the allowable energy levels of the electrons passing through it resulting in a reduction of noise. Also, by maintaining sufficient spacing between the energy levels in the quantum barrier, scattering due to optical phonons is minimized, further reducing noise. The reduction in base width will also result in a large gain as is seen in NBHBTs.

Device structure and modeling

All modeling was performed using the commercial simulation program Silvaco ATLAS. First, modeling of a commercially available HBT was performed to provide a starting point. This first model was adjusted so that the modeled results were a close match to the measured results of the manufactured device. This HBT has an emitter region formed of highly n-type doped cap layers of InGaAs at the emitter contact followed by a region of n-type GaAs and then a region of n-type InGaP. The base includes 10 ¹⁹ cm ^"3 p-type doped GaAs of a thickness of 1000 A . The collector includes n-type GaAs. The device introduced in this example is referred to as a quantum well base heteroj unction bipolar transistor (QWBHBT). A quantum well was created in the base of the structure by creating a double heteroj uction. The base material is still formed of 10 ¹⁸ cm ^"3 p-type doped GaAs, but the thickness of this region is reduced to 240A so that it is comparable to the wavelength of the electrons passing through it. The InGaP region in the emitter, adjacent to the base, remains. An additional InGaP region in the collector adjacent to the base was added. The thickness of both InGaP regions is 400 A and both regions are n-type doped. The combination of the lower energy gap p-type GaAs region sandwiched between two higher energy gap n-type InGaP regions creates a double heterojunction that results in the formation of a deep quantum well for the holes injected into the base from the base contact and a quantum barrier for the electrons injected from the emitter through the base into the collector. The remaining structure remains the same as in the commercial device. The QWBHBT structure can be seen in Fig. 5. The width of both modeled devices was 38.3 μm.

Results and discussion

A set of collector current vs. collector voltage curves (I-V curves) was found using the modeling software for both the commercial HBT and the QWBHBT. Fig. 6 shows the I-V curves for the commercial HBT. In comparison, Fig. 7 shows the output characteristic of the QWBHBT. A significant improvement in gain can be seen with the quantum well base device where there is an increase by a factor of 10 from 75 to 750. This improvement in β is mainly due to the reduction in base thickness that results in a reduced recombination current across the thin neutral base. There are other notable features in the IV curve comparison. First the DC gain decreases as the base current increases. This effect becomes more severe at higher base current values where β approaches zero at a base current of 300 μA. The phenomenon of losing control of the transistor operation is linked to the appearance in the base of a large number of free electrons that are screening out the depleted region of the collector junction. The base area at this point behaves not as a regular doped semiconductor but rather as a semi-metal. Additionally, a region of negative resistance occurs at higher base currents and at lower collector voltages. At high output currents, a large number of electrons experience a change of mobility coming from the emitter material to the base material where mobility is lower. AC gain in the form of the S ₂ i parameter vs. frequency for a base current of 50 μA and a collector bias of 2.5V at an emitter bias of zero was also found. Although the region of the collector adjacent to the base was changed to InGaP in the QWBHBT device as compared to the structure of the commercial HBT, the collector doping was kept the same. Hence, the depletion region width at the base collector junction remains unchanged. In addition to this, the reduced base thickness causes a reduction in base transit time. This can be seen in the frequency response of the transistor. Fig. 8 shows the S ₂₁ in decibels plotted against frequency for both the QWBHBT and the commercial HBT. The plot shows that the QWBHBT device does not just possess a larger DC current gain but also an improvement of 5 dB of AC gain in the frequency range of 1-10 GHz. The simulation also predicts that the cutoff frequency of the QWBHBT is about 50GHz.

Gummel plots were found for both devices and were taken at a collector bias of 2.5V while the base bias was swept and the emitter was held at ground. The comparison can be seen in Fig. 9. The large gain of the QWBHBT can be seen in comparison to that of the commercial HBT.

Figs. 10 and 11 show the change in the current gain with respect to the increase in the collector current for both of the transistors. The commercial HBT shows a linearity of around 0.3 dB at collector currents of 20-4OmA with an average DC gain of 37.7 dB, while the QWBHBT shows a linearity of around 0.4 dB at collector currents of 20-8OmA with an average DC gain of 57 dB. Fig. 10 also shows how the gain starts decreasing as the output current increases for the QWBHBT.

Figs. 12 and 13 show the modeling results at a temperature of 150 K. In Fig. 12, the commercial HBT results are unremarkable while the QWBHBT curves show distinctive plateaus indicative of quantization in the base of the device. Similar results have been observed and measured in other manufactured quantum devices. It should be noted that when the quantum simulator is not employed, that these plateaus are not present in the resulting I- V curves.

Discussion The development of energy gap engineering led to the design of HBTs where the heterostructure significantly increased emitter efficiency. The ability to grow epitaxial layers with a high precision of up to one atomic layer opened the way to create HBTs with quantum wells where resonant tunneling was used to create negative resistance. Quantum wells in the base of HBTs have not been used before for the enhancement of amplifier performance. In the current study, we modeled a HBT with a quantum well base to demonstrate significant improvements of transistor gain, power, and operational frequency. The precise control of electron transport through the HBT, i.e., electron wavelength engineering, is expected to further improve the performance of bipolar transistors with regard to linearity and noise level.

Having thus described an illustrative embodiment of the invention, it should be apparent to those skilled in the art that the foregoing is merely illustrative and not limiting, having been presented by way of example only. Numerous modifications and other illustrative embodiments may be contemplated by those of ordinary skill in the art and are believed to fall within the scope of the invention. For example, although a QEBHBT transistor may be formed using materials such as GaAs and InGaP as described herein, other suitable materials may be used, as the techniques described herein are not limited to particular materials. For example, other types of III-V semiconductor materials may be used.

What is claimed is: