APPARATUS AND METHOD OF FORMING A MOSFET WITH ATOMIC LAYER DEPOSITED GATE DIELECTRIC

PRIORITY CLAIM This Application claims priority to United States Provisional Patent

Application Serial No. 60/809,195 filed on May 30, 2006, the entirety of which is incorporated by reference herein.

FIELD OF THE DISCLOSURE The present invention relates generally to metal oxide semiconductor field-effect transistors (MOSFETs), and more specifically to enhancement mode MOSFETs.

BACKGROUND Innovative device structures and gate dielectrics allow the advancement of developing Si complementary metal oxide semiconductor (CMOS) integration, functional density, speed and power dissipation, and extend CMOS front- end fabrication to and beyond the 22 -nm node. One emerging strategy is to use IH-V compound semiconductors as conduction channels, to replace traditional Si or strained Si, while integrating these high mobility materials with novel dielectrics and heterogeneously integrating them on Si or silicon-on-insulator (SOI). For more than four decades, the research community has been searching for suitable gate dielectrics or passivation layers on III-V compound semiconductors. One obstacle is the lack of high-quality, thermodynamically stable insulators on materials such as GaAs that can match the device criteria as SiO ₂ on Si, e.g., a mid-bandgap interface-trap density (Dn) of ~10 ¹⁰ /cm ² -eV. Recently, in situ molecular beam epitaxy (MBE) growth of Ga ₂ θ ₃ (Gd ₂ θ ₃ ) and ex situ atomic layer deposition (ALD) growth of Al ₂ θ ₃ attract particular attention. Research involving ALD high-k dielectrics is of particular interest, since the Si industry is getting familiar with ALD Hf-based dielectrics and this approach has the potential to become a manufacturable technology.

SUMMARY

According to one aspect of the disclosure, a method of forming a metal oxide semiconductor field-effect transistor (MOSFET) includes forming a III-V compound semiconductor on a substrate with the IH-V compound semiconductor being doped with a first dopant type. The method further includes doping a first and second region of the III-V compound semiconductor with a second dopant type to form a drain and a source of the MOSFET. The method further includes forming a gate dielectric on the III-V compound semiconductor through atomic layer deposition. The method further includes applying a metal onto the dielectric to form a gate of the MOSFET.

According to another aspect of the disclosure, a MOSFET includes a substrate and a III-V compound semiconductor formed on the substrate and being doped with a first dopant type. The III-V compound semiconductor includes a first region doped with a second dopant type to form a drain of the MOSFET and a second region doped with the second dopant type to form a source of the MOSFET. The MOSFET further includes a gate dielectric formed on the III-V compound semiconductor through atomic layer deposition. The MOSFET further includes a gate formed of metal disposed on the gate dielectric.

According to another aspect of the disclosure, a MOSFET includes a substrate and a III-V compound semiconductor formed on the substrate. The III-V compound semiconductor is doped with a first dopant type. The III-V semiconductor includes a drain region and source region each doped with a second dopant type. The MOSFET further includes a gate dielectric formed on the III-V compound semiconductor through atomic layer deposition. The MOSFET further includes a gate formed of metal on the gate dielectric.

BRIEF DESCRIPTION OF THE DRAWINGS

The detailed description particularly refers to the accompanying figures in which: FIG. l(a) is a cross-sectional view of an illustrative metal-oxide semiconductor field-effect transistor (MOSFET) and an illustrative capacitor formed on the same wafer;

FIG. l(b) is a plot of leakage current density versus gate bias of an illustrative MOSFET;

FIG. 2(a) is a plot of capacitance versus voltage of an illustrative MOS capacitor; FIG. 2(b) is a plot of hysteresis versus frequency of an illustrative

MOS capacitor;

FIG. 2(c) is a plot of accumulation capacitance versus frequency of an illustrative MOS capacitor;

FIG. 2(d) is a plot of flat band voltage versus frequency of an illustrative MOS capacitor;

FIG. 3 (a) is a plot of capacitance versus voltage of an illustrative MOSFET;

FIG. 3(b) is a plot of current versus voltage of an illustrative MOSFET; FIG. 4(a) is a cross-sectional view of another illustrative MOSFET;

FIG. 4(b) is a plot of drain-source current versus applied bias for an illustrative MOSFET;

FIG. 5(a) is a plot of channel resistance versus mask designed gate length of an illustrative MOSFET; FIG. 5(b) is a plot of drain-source current versus gate voltage for an illustrative MOSFET;

FIG. 5(c) is a plot of drain-source current versus the inverse of mask designed gate length of an illustrative MOSFET;

FIG. 6(a) is a plot of capacitance versus voltage of an illustrative MOS capacitor; and

FIG. 6(b) is a plot of carrier mobility versus electric field for an illustrative MOSFET.

DETAILED DESCRIPTION OF THE DRAWINGS First Illustrative Embodiment

Referring now to FIG. l(a), there is a cross-section of one illustrative embodiment of a device structure of a fabricated enhancement mode (E-mode) n- channel metal-oxide semiconductor field-effect transistor (MOSFET) 10

implementing a III-V compound semiconductor as a channel layer. In particular, FIG. 1 shows an illustrative embodiment of the MOSFET 10 implementing Ino.2Gao.8As as the channel layer 12 with AI ₂ O ₃ serving as a high-k gate dielectric layer 14. A high-k dielectric such as AI2O ₃ , can provide low defect density, low gate leakage, and high thermal stability. In this illustrative embodiment, the gate dielectric 14 is integrated on the channel layer 12 through atomic layer deposition (ALD).

In the illustrative device structure shown in FIG. l(a), the MOSFET 10 further includes a buffer layer 16 and an intermediate layer 18. In the illustrative embodiment shown in FIG. 1, the channel layer 12, the buffer layer 16, and the intermediate layer 18 were sequentially grown by metal-organic chemical vapor deposition (MOCVD) on a substrate 20, shown in this illustrative embodiment as 2- inch GaAs p-h In one illustrative embodiment, the buffer layer 16 is a 150 nm p- doped 4xlO ¹⁷ /cm ³ Ino. ₂ Gao. ₈ As layer, the intermediate layer 18 is a 285 nm p-doped IxIO ¹⁷ /cm ³ Ino. ₂ Gao. ₈ As layer, and the channel layer 12 is a 13.5 nm p-doped IxIO ¹⁷ /cm ³ Ino.2Gao.8As channel layer. It should be appreciated that the methods described herein enable one of ordinary skill in the art to form various III-V compound semiconductors on compatible substrates that allow MOSFETs, such as the MOSFET 10, to be formed. Returning to the MOSFET 10, it should be appreciated that the doping concentrations of the layers 12, 16, and 18 can vary in range, such as in the ' order of 10 ¹⁶ to 10 ¹⁸ , for example. In one illustrative embodiment, after appropriate surface pretreatment, 16 nm — 30 nm ALD AI2O ₃ can be deposited at approximately 300 ⁰ C through atomic layer deposition. In one illustrative embodiment the ALD was performed using an ASM Pulsar2000™ ALD module, however it should be appreciated other devices can be used to perform the atomic layer deposition. Compared to the conventional methods to form thin AI2O ₃ films, i.e., by sputtering, electron beam evaporation, chemical vapor deposition or oxidation of pure Al films, the ALD AI ₂ O ₃ is typically of a much higher quality. ALD is an ultra- thin-film deposition technique based on sequences of self-limiting surface reactions enabling thickness control on atomic scale. The ALD high-k materials including AI2O3 can be used to substitute SiO ₂ for sub-100 nm Si complementary MOSFET applications, as is conventionally used. ALD also allows the integration of high- quality gate dielectrics on non-Si semiconductor materials such as Ge, a high mobility

In the illustrative device structure of FIG. l(a), the ALD Al ₂ O ₃ gate dielectric 14 serves not only as a gate dielectric but also as an encapsulation layer due to its high thermal and chemical stability. In this illustrative embodiment, the source and drain regions 22, 24 are Si implanted that may be activated at 750 - 850 ⁰ C by rapid thermal annealing (RTA) in N ₂ ambient. Dopant activation annealing is a step that not only activates the dopant, but may also preserve the smoothness of the interface at the atomic level. Using a wet etching in diluted hydrofluoric acid (HF) in the illustrative MOSFET 10, the oxide on the source and drain regions 22, 24 was removed, while the gate area 26 was protected by photoresist. It should be appreciated that the doping of a III-V compound semiconductor such as that disclosed regarding the MOSFET 10 can be varied. For example, the illustrative MOSFET 10 includes a p-type dopant in the channel layer 12 and an n-type dopant in the source 22 and drain regions 22, 24. However, in other embodiments, the channel layer 12 may include an n-type dopant and the source and drain regions 22, 24 may include a p-type dopant. This allows MOSFETs, such as the MOSFET 10, to be implemented for PMOS and NMOS configurations.

In one illustrative embodiment, ohmic contacts 28 were formed by electron-beam deposition of Au/Ge/Au/Ni/Au and a lift-off process, followed by an approximate 400 ⁰ C anneal in N ₂ ambient. Conventional Ti/Au metals were e-beam evaporated, followed by lift-off to form the gate electrode 27. This illustrative process includes 4 levels of lithography (alignment, source and drain implantation, ohmic, and gate), which may all be done using a contact printer. The sheet resistance of the implanted source/drain regions 22, 24 and their contact resistances may be measured from transfer length method (TLM) to be ~ 300 ω/sq. and ~ 1.0 ω ^* mm, which, in this illustrative embodiment, demonstrates good process on implantation and activation. The designed gate lengths in various illustrative embodiments of the MOSFET 10 are 0.65, 0.85, 1, 2, 4, 8, 20 and 40 μm. It is not a self-aligned process. The overlap length between gate area 26 and source/drain 22, 24 is estimated around ~ 0.5 μm or less. Also shown in FIG. l(a) is a capacitor 32, which is formed on the same wafer as the MOSFET 10. It should be appreciated that the capacitor 32 can be used for purposes of determining expected electrical characteristics of the MOSFET 10 and that in most nractical annlicationς the MOSFET 10 would be implemented separately.

Al ₂ O ₃ dielectric films are highly electrically insulating. In this illustrative embodiment, the AI ₂ O ₃ gate dielectric 14 shows very low leakage current- density of ~10 ^'9 to 10 ^~8 A/cm ² for amorphous films thicker than 6 nm. The leakage current density starts to increase after high temperature annealing, which is required to activate Si dopants implanted in GaAs. The increase of leakage currents is due to the creation of more leakage paths around crystallized grains in the amorphous films after high temperature annealing. As shown in FIG. l(b), the leakage current density (Ji) is measured on 30 nm ALD Al ₂ O ₃ versus the applied potential on the capacitor 32, gate bias V _g , with high annealing temperatures. The positive bias means that the metal electrode 29 is positive with respective to GaAs as shown in Fig. l(a). The plot of FIG. l(b) shows extremely low leakage current density in ALD AI ₂ O ₃ films even after 800 ⁰ C (approximately) annealing, though a significant increase in current density exhibits with increasing annealing temperature up to approximately 820 ⁰ C. An approximate temperature of 85O ⁰ C is the critical temperature when the crystallization in amorphous AI ₂ O ₃ becomes severe and the leakage current increases dramatically.

In one illustrative example, the high-quality OfAl ₂ O ₃ ZInGaAs interface surviving from high temperature annealing is verified by capacitance-voltage (C-V) curves showing sharp transition from depletion to accumulation with "zero" hysteresis, 1% frequency dispersion per decade at accumulation capacitance and strong inversion at split C-V measurement. (See FIGS. 2(a) - 3(b).) For purposes of this disclosure, the term "split" refers to one terminal split into three, such as a source, drain, and back gate with all three ground together. Further, ALD AI ₂ O ₃ allows an (E-mode) n-channel III- V MOSFET to have a true inversion channel formed at the high-k dielectric interface, such as the Al ₂ O ₃ /InGaAs interface of the MOSFET 10 shown in FIG. 1.

Illustrative C-V measurements on high temperature annealed ALD Al ₂ O ₃ dielectrics on InGaAs and I-V (current-voltage) characterization on an illustrative fabricated E-mode InGaAs MOSFET 10 where the inversion channel is directly formed at the AIaO ₃ ZInGaAs interface are shown in FIGS. 3 (a) and 3(b). Al ₂ O ₃ may be used as an insulating material for a gate dielectric based upon its tunneling barrier and protection coating due to its excellent dielectric properties, strone adhesion to dissimilar materials, and thermal and chemical stability. Al ₂ O ₃

typically has a high handgap (~9 eV), a high breakdown electric field ( approximately 5-30 MV/cm), a high permittivity (approximately 8.6-10) and high thermal stability (up to at least 1000 ⁰ C) and remains amorphous under typical processing conditions for implanted dopant activation on GaAs. It should be appreciated that other materials can be used as a dielectric in the MOSFET 10, such as HfO ₂ , ZrCb, Ga ₂ Oa, Gd ₂ O ₃ , Y ₂ O ₃ , TiO ₂ , Ta ₂ O ₅ , La ₂ O ₃ , and their combinations such as HfAlO, TiAlO, and LaAlO. It should be further appreciated that a gate dielectric could be formed of SiO ₂ , Si ₃ N ₄ formed by CVD (chemical vapor deposition) or PVD (physical vapor deposition) as an encapsulation layer. C-V measurements allow a quantitative study of a MOS structure.

From the C-V measurements, three quantities allowing the evaluation of high-k dielectrics on novel channel materials may be determined. The first is the amount of hysteresis that results when the MOS capacitor is biased well into accumulation and inversion. The second is the interface trap density Dj _t at the interface showing in C-V curve how rapid the transition is between accumulation and inversion. The third is the frequency dispersion on accumulation capacitances and flat-band shifts. The C-V characterization of ALD AI2O3 on InGaAs may be observed after high temperature annealing, such as between 750 ⁰ C — 850 ⁰ C for example, which is required to activate Si dopants in InGaAs. It should be appreciated that improved C-V curves or Du may be exhibited under annealing temperatures, such as those between 45O ⁰ C — 600 ⁰ C for example, under a certain ambient or complicated annealing process. However, these curves are not directly relevant to the MOS interface needed for realizing inversion- channel E-mode GaAs MOSFET using implantation process.

The high-frequency (10 kHz) C-V curves for the illustrative capacitor 32 with a 30 nm ALD Al ₂ O ₃ on InGaAs is shown in FIG. 2(a). This illustrative capacitor 32 experienced all the various device processes previously described herein and was annealed with an RTA step of approximately 800 ⁰ C for approximately 10 seconds in nitrogen ambient. In this illustrative embodiment, the solid curve in FIG. 2(a) is measured from -5 V to +3 V with the sweep rate ~ 4V/min., while the dashed curve is taken from +3 V to -5 V. The solid and dashed curves indicate substantially "zero" hysteresis in this C-V loop with the maximum shift less than 20 raV.

FIG. 2(b) illustrates the hysteresis versus frequency observed on a typical MOS capacitor, such as the MOS capacitor 32. The hysteresis of FIG. 2(b) is

in the range of 10 - 50 mV between 1 KHz - 1 MHz ₅ and increases as the frequency goes down in general. The extremely low hysteresis illustratively shown in FIG. 2(b), along with the sharp transition from depletion and accumulation, indicates the high quality of bulk and interface properties of ALD AI2O ₃ on InGaAs even after high temperature annealing. The effects of the annealing step on the chemical and structural properties of the interface are very complex and outside the scope of this disclosure.

It should be appreciated that the frequency dispersion on accumulation capacitance C _max is another issue for consideration regarding high-k dielectrics on III- V materials. In one illustrative embodiment of the MOSFET 10, this dispersion could be as large as 50% or more in the frequency range of 1 kHz to 1 MHz, which stymies all efforts to estimate Du using high-low frequency capacitance method. FIG. 2(c) illustratively summarizes the accumulation capacitance C _max measured on exemplary 800 ⁰ C annealed capacitors 32 in the wide frequency range from 500 Hz up to 1 MHz. The frequency dispersion is only 1 % per decade at this frequency range. This experiment illustratively demonstrates that the major part of frequency dispersion on a non-ideal oxide/III-V material interface does relate to the interface properties instead of simple parasitic effect, which may be corrected by two-frequency correction or multi circuit element models. The dielectric constant is ~ 8.1 deduced from the measured maximum accumulation capacitance C _max , the area of the capacitor 32, and the dielectric film thickness.

The flat band shift is also an issue of interest at the beginning for alternative dielectrics research on Si. Of note is the observed frequency dependent flat band shift of 800 ⁰ C annealed capacitors. FIG. 2(d) illustratively shows the summary of the flat band voltage V _n , versus frequency on exemplary 16 nm and 30 ran thick ALD Al ₂ O ₃ films. In this illustrative example, the frequency dependent flat band shift is much less on medium temperature (45O ⁰ C — 600 ⁰ C) annealed capacitors 32. This phenomenon is less significant on 16 nm thick film compared to 30 nm thick one. It is roughly scaled with the film thickness and linearly dependent on log(frequency).

In one illustrative embodiment, clear n-channel inversion on p-type InGaAs is realized at AbOs/InGaAs interface, which is demonstrated by measuring Cgbc, the capacitance measured from the gate 118 when the source 108, drain 110, and

back gate (not shown) are all grounded, (via split-C-V method) on MOSFETs, such as the MOSFET 10, as illustratively shown in FIG. 3(a). The C-V curve is taken on an ^' exemplary MOSFET 10 with a 40 μm gate length and a 100 μm gate width at frequency as high as 1 kHz. At the conventional MOS configuration as measured in FIG. 2(a), no n-channel inversion is observed at frequencies as low as 100 Hz. It should be appreciated that the minority carriers (electrons) are provided by the source and drain diffusions and not by thermal generation in the depletion region. In the split C-V or MOSFET configuration with source and drain grounded at the same time, majority (holes) and minority (electrons) carrier capacitances may be measured independently at the same frequency. In one illustrative embodiment, using a low- frequency capacitance of C _LF ⁼ 5.5 pF, a high-frequency capacitance of C _HF ⁼ 5.0 pF, and an oxide capacitance of C _0x -11.0 pF, the middle gap interface trap density Dj _t is determined to be 2.9 x 10 ^u /cm ² -eV.

FIG. 3(b) shows the I-V characteristics of an illustrative 1 x 100 μm ² gate geometry E-mode n-channel InGaAs MOSFET 10 with ALD Al ₂ O ₃ as a gate dielectric, such as that shown in FIG. l(a). In this illustrative example, the gate voltage is varied from 12 to 0 V in steps of -2 V and the threshold voltage, V _T , is ~ 0. It is believed that the low drain current could be improved by optimizing implant and annealing conditions to reduce parasitic resistance and surface scattering. The maximum drain current on the illustrative MOSFET 10 device is ~0.12 mA/mm, which is comparable to the GaAs based E-mode MOSFET 10 reported known at Ga ₂ θ ₃ (Gd2θ ₃ )/GaAs interface. It should be appreciated that by combining the C-V result in FIG. 3(a) and the I-V result in FIG. 3(b), it is clearly demonstrated that the true inversion n-channel can be formed at ALD AIaO ₃ ZInGaAs interface.

Second Illustrative Embodiment

FIG. 4(a) shows the cross-sectional diagram of an ALD

Al ₂ 0 ₃ /Ino.s3Gao.47As MOSFET 100. It should be appreciated that, based upon the descriptions of both MOSFETs 10, 100 described herein, that HI-V compound semiconductors, for purposes of this disclosure, can include varied amounts of a particular element with respect to the other elements in the compound. For example, In can elementally make up about 15% - 100% of the HI-V compound semiconductors used in forming MOSFETs 10, 100. Returning to the MOSFET 100 _>

a 500 run p-doped 4xlO ¹⁷ cm ^"3 buffer layer 102, and a 300 run p-doped IxIO ¹⁷ cm ^"3 Ino.s ₃ Gao. ₄₇ As channel layer 104 that were sequentially grown by MBE on a 2-inch InP p+ substrate 106. In one illustrative embodiment, after surface degreasing and ammonia-based native oxide etching, wafers including the buffer layer 102, channel layer 104, and substrate 106 were transferred via room ambient to an ASM F- 120 ALD reactor. In this embodiment, a 30 nm thick AI ₂ O3 dielectric layer 103 was deposited at a substrate temperature of approximately 300 ⁰ C as an encapsulation layer. A source region 108 and a drain region 110 were selectively implanted with a Si dose of IxIO ¹⁴ cm ^"2 at 30 keV and IxIO ¹⁴ cm ^"2 at 80 keV through the 30 nm thick AI ₂ O ₃ layer. Implantation activation was achieved by RTA at approximately 650 — 850 ⁰ C for 10 seconds in a nitrogen ambient. An 8-nm AI ₂ O ₃ film was regrown by ALD after removing the encapsulation layer by BOE solution and ammonia sulfide surface preparation. After approximately 600 ⁰ C PDA process, source and drain ohmic contacts 114 were made by an electron beam evaporation of a combination of AuGe/Ni/Au and a lift-off process, followed by a RTA process at approximately 400 ⁰ C for 30 seconds also in a N ₂ ambient. A gate electrode 118 was defined by electron beam evaporation of Ni/ Au and a lift-off process. The fabricated MOSFETs 100 have a nominal gate length varying from approximately 0.50 μm to 40 μm and a gate width (not shown) of approximately 100 μm. In one illustrative embodiment, an HP4284 LCR meter was used for the capacitance measurement and a Keithley 4200 was used for output characteristics of the MOSFET 100.

FIG. 4(b) shows the dc Ia _s (drain current) - V _ds (aρplied bias) characteristics of the MOSFET 100 for one illustrative embodiment with a gate bias from 0 to 5 V in steps of +1 V. The characteristics shown in FIG. 4(b) are for an embodiment of a MOSFET 100 having a mask designed gate length L^ _ask of approximately 0.50 μm and gate width of approximately 100 μm. In one illustrative embodiment, a maximum drain current of 367 mA/mm is obtained at a gate bias of 5 V and a drain bias of 2 V. In this illustrative embodiment, the MOSFET 100 performance has a significant leap with 3000 times increase of the maximum drain current, compared to the results discussed in regard to the first illustrative embodiment implementing In _o .2oGao. ₈ oAs, such as the MOSFET 10. In one illustrative embodiment, the gate leakage current is below 1.4 μA/mm under the same bias condition used in regard to the MOSFET 10, more than five orders of magnitude

bias condition used in regard to the MOSFET 10, more than five orders of magnitude smaller than the drain on-current. The source-drain leakage current is another issue for narrow bandgap semiconductor devices caused mainly by drain-induced-barrier- lowing (DIBL) effect and impact ionization. As shown in FIG. 4(b), the source-drain leakage current at zero gate bias is between approximately 1 uA to 100 μA at V _< i _s =2 V.

Since the fabrication process used is typically not a self-aligned process, accurate determination of the effective gate length and series resistance is important in evaluation of the intrinsic device performance and the potential for further optimization. FIG. 5(a) shows an illustrative effective gate length L _e ff and a series resistance R _SD extracted by plotting measured channel resistance Rc _h vs. Here, Rc _h is the measured channel resistance and Lua _s k is the mask designed gate length. The series resistance R _SD and the difference between ∑Mask and L _e ff, designated as AL , are determined to be 15 ω and <0.05 μm, respectively. AL is negligible in this L. In this embodiment, a maximum extrinsic transconductance G _m is approximately 130 mS/mm and on-resistance is only 2 ω-mxn at V _g =SV. The extrinsic transconductance G _m could be further improved by reducing the thickness of the dielectric and improving the quality of the interface. To evaluate the output characteristics more accurately, the intrinsic transfer characteristics are calculated by subtracting half of the series resistance R _SD and are compared with extrinsic ones in Fig. 5(b). The resulting intrinsic maximum drain current Id _s and transconductance G _m for 0.5 μm device are 425 rnA/mrn and 145 mS/mm, respectively. By the conventional linear region extrapolation method or second derivative method, the extrinsic threshold voltage is determined around 0.25 V in one illustrative embodiment. The threshold voltage can be finely tuned by channel doping and metal work function to fulfill the specific requirements for different applications. The subthreshold slope and DIBL for one particular embodiment of the MOSFET 100 are 260 mV/dec and 400 mV/V, respectively.

Fig. 5(c) summarizes all measured drain current Ids vs. \/L _maS k under and V _ds =2V or the illustrative MOSFET 100. The drain current or transconductance G _n , is linearly inversely proportional to the mask designed gate length L _mask as expected and start to saturate at approximately L _mas k ⁼ 0.75 μm. In other illustrative embodiments, the surface preparation and high-k dielectric formation

process may be optimized, such that the drain current and transconductance G _m could increase at least a factor of 2—3 further.

As similarly shown in regard to the MOSFET 10, detailed C-V measurements of MOS capacitors, fabricated on the same device wafers, may also carried out to evaluate the interface quality of ALD AI ₂ O ₃ on Ino. ₅₃ Gao.4 ₇ As as illustratively shown in Fig. 6(a). The high-frequency (HF) C-V Ά\ 10 kHz shows a clear transition from accumulation to depletion for a typical p-type MOS capacitor. The inversion features of low-frequency (LF) C- V from 1 kHz down to 300 Hz indicate that the conventional Fermi-level pinning phenomenon on III-V is overcome in this ALD-Al ₂ θ ₃ /Ino.53Gao. ₄₇ As interface. It is believed that the small (3% per decade) frequency dispersion at accumulation capacitance to the relative high Du at the conduction band edge, though the extrinsic parasitic effects could also contribute to it. The mid-gap Du is estimated to be around 1.4 x 10 ¹² /cm ² -eV determined by HF- LF method. Moderate hysteresis of 100-300 mV exhibits in the C-V loops (not shown).

Effective mobility is another important parameter to evaluate the MOSFET 100 performance. A split C-V method is used to measure the channel capacitance of a 40-μm- gate-length device which can be used to calculate the total inversion charge in the channel by integrating the C-V curve. The extracted effective mobility μ _ej y has a peak value of 1100 cm ² /Vs around a normal electric field E _e ffθf

0.25 MV/cm and twice higher μ _ej p than Si universal mobility at E _ej yof 0.50 MV/cm as shown in Fig. 6(b). A better mobility performance is expected to be achievable by further optimizing dielectric formation and device fabrication process.

There are a plurality of advantages of the present disclosure arising from the various features of the apparatus and methods described herein. It will be noted that alternative embodiments of the apparatus and methods of the present disclosure may not include all of the features described yet still benefit from at least some of the advantages of such features. Those of ordinary skill in the art may readily devise their own implementations of an apparatus and method that incorporate one or more of the features of the present disclosure and fall within the spirit and scope of the present disclosure.