DIELECTRIC BARRIER LAYER

Abstract

The present invention relates to a method of forming a fluorine-doped metal oxide dielectric layer suitable for forming a dielectric barrier layer in an integrated circuit device. The method comprises the deposition of a plurality of layers of oxide dielectric onto a substrate by a plurality of cycles of atomic layer deposition, wherein one or more of said cycles of atomic layer deposition additionally comprises the atomic layer deposition of fluorine. In addition, the present invention relates to the dielectric films formed by this methodology and to integrated electronic devices that comprise these metal oxide dielectric barrier layers.

Claims

1. A method of forming a fluorine-doped metal oxide dielectric layer suitable for forming a dielectric barrier layer in an integrated circuit device, the method comprising the deposition of a plurality of layers of metal oxide dielectric onto a substrate by a plurality of cycles of atomic layer deposition, wherein one or more of said cycles of atomic layer deposition additionally comprises the atomic layer deposition of fluorine.

2. A method according to claim 1, wherein the substrate is a layer of a gallium nitride (GaN) layer, AlN, AlGaN layer, AlInN layer or nitride based alloy in which a two dimensional electron/hole gas can be formed as the channel.

3. A method according to claim 1, wherein the oxide dielectric is selected from alumina (Al.sub.2O.sub.3); hafnia (HfO.sub.2); zirconia (ZrO.sub.2); titania (TiO.sub.2); rare earth element (RE) oxides (RE.sub.2O.sub.3 or REO.sub.2); or a mixture thereof.

4. A method according to claim 3, wherein the oxide dielectric is in a pure state.

5. A method according to claim 3, wherein the oxide dielectric is in a doped state.

6. A method according to claim 3, wherein the oxide dielectric is a ternary or quaternary composition of the metal oxide dielectric, which are either in a pure or doped compositional state.

7. A method according to claim 1, wherein the oxide dielectric is alumina.

8. A method according to claim 7, wherein alumina is deposited by the atomic layer deposition of an aluminium source (e.g. trimethyl aluminium) followed by the either the atomic layer deposition of water vapour, oxygen plasma treatment or ozone treatment.

9. A method according to claim 1, wherein the thickness of the oxide dielectric layer is from 1 nm to 500 nm.

10. A method according to claim 1, wherein the total number of cycles of atomic layer depositions may be within the range of 5 to 100,000 cycles.

11. A method according to claim 10, wherein the total number of cycles of atomic layer depositions may be within the range of 50 to 2000 cycles.

12. A method according to claim 1, wherein fluorine is deposited by introducing a fluorine source between doses of the metal oxide source.

13. A method according to claim 1, wherein fluorine is introduced in one of every 1 to 200 cycles of oxide deposition.

14. A method according to claim 1, wherein fluorine is introduced in a regular manner so as to provide a regular distribution of fluorine throughout the metal oxide dielectric layer.

15. A method according to claim 1, wherein fluorine is introduced in an irregular manner to provide a more varied distribution of fluorine throughout the metal oxide dielectric barrier layer.

16. A method according to claim 1, wherein the fluorine is introduced by the introduction of a fluorine source into the ALD apparatus and said fluorine source is selected from the group consisting of fluorocarbons that are miscible with water or other suitable solvent, xenon fluoride, or an aqueous fluoride solution.

17. A method according to claim 16, wherein the fluoride source is an aqueous solution of ammonium fluoride.

18. A method according to claim 1, wherein fluorine is deposited in the presence of a co-dopant.

19.-23. (canceled)

24. A fluorine-doped metal oxide dielectric barrier layer obtainable by a method according to claim 1.

25. (canceled)

26. An integrated circuit device comprising a fluorine-doped metal oxide dielectric barrier layer according to claim 24.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The invention is described further in reference to the accompanying Figures in which:

[0057] FIG. 1a shows a depth profile of the fluorine distribution in the 10% and 100% cycle doped films obtained using secondary ion mass spectrometry;

[0058] FIG. 1b shows a plot of n.sup.2−1 versus 1/λ.sup.2 for spectroscopic ellipsometry measurements for films of undoped alumina and with doping cycle percentages of 10% and 100%;

[0059] FIGS. 1c and 1d show transmission electron micrographs recorded from cross sections of MOS capacitors fabricated from the Al.sub.2O.sub.3:F films;

[0060] FIG. 2 shows the following:

[0061] FIG. 2(a) Capacitance-voltage curves show F-addition shifts the threshold voltage and reduces the hysteresis caused by defects in the oxide

[0062] FIG. 2(b) dC/dV makes the positive flatband voltage due to F-addition much more apparent

[0063] FIG. 2(c) The leakage current (I-V) data shows leakage suppression although 10% seems to reinstate leakage again—suggesting upper limit to benefit from doping

[0064] FIG. 2(d) All samples benefit from forming gas annealing (FGA) at 430° C. for 30 minutes

[0065] FIGS. 3a to 3f show the transistor characteristic of drain current under conditions of Drain-Source (DS) and Gate-Source (GS) voltage bias.

EXAMPLES

Example 1—the In-Situ ALD Fluorine Doping Process

[0066] In the following description of the methodology of the present invention, it will be appreciated that the cyclic nature of the introduction of the organometallic and oxidant co-reagent (either H.sub.2O or NH.sub.4F/H.sub.2O) means that the fluorine content of the dielectric can be controlled by varying the ratio of H.sub.2O:NH.sub.4F/H.sub.2O cycles used.

[0067] The ALD Al.sub.2O.sub.3 dielectric films were deposited onto GaN on Si(111) wafers grown by Metal Organic Chemical Vapor Phase Deposition (MOCVD). The three precursors (SAFC Hitech) were trimethyl aluminium (TMA), deionised water and a 16% or 40% aqueous solution of ammonium fluoride.

[0068] The atomic layer deposition experiments were performed in an Oxford Instruments Plasma Technology OpAL Plasma ALD reactor incorporating a 200 mm diameter heater controllable between 25 and 500° C. The system is pumped by an oil filled rotary pump, capable of achieving a base pressure of approximately 20 mTorr. The liquid TMA, H.sub.2O and 16% or 40% NH.sub.3F/H.sub.2O sources were delivered, at ambient temperature, into the process chamber from independent sources. The precursor delivery was achieved via vapour draw. The precursor delivery lines from the module to the process chamber are heated. Precursors are delivered into the chamber by fast ALD valves with a minimum opening time of 1 ms. Details of the ALD parameters employed are shown in table 1 below.

TABLE-US-00001 TABLE 1 ALD growth parameters Substrate Temperature 200° C. Pressure 200 mTorr Argon flow 100 std cm.sup.3min.sup.−1 Pulse Sequences: TMA [0.02 s dose/2 s purge]  H.sub.2O [0.3 s dose/2 purge] .sup.  16% NH.sub.3F/H.sub.20 [0.3 s dose/2 s purge]

[0069] The Al.sub.2O.sub.3:F films are deposited using x cycles of Al.sub.2O.sub.3 steps, via exposure of the surface to successive steps of TMA and then water vapour. Intermittent pump purge steps are used to prevent gas-phase pre-reaction, which ensures the surface reaction. After the x [Al.sub.2O.sub.3]cycles, a single Al.sub.2O.sub.3:F cycle is deposited, via exposure of the surface to successive steps of TMA and then ammonium fluoride solution. The whole process is repeated N times until the required film thickness is achieved. A typical growth run would use a Al.sub.2O.sub.3:F ratio of 10:1 whereby x=10 and N=100, giving a total number of 1000 Al.sub.2O.sub.3 and 100 Al.sub.2O.sub.3:F ALD cycles. The temperature-independent growth “plateau” occurs over a temperature range of 185° C. to 300° C. and denotes the deposition range where the enthalpy of surface reactions dominates the deposition rate, rather than the substrate thermal energy. Above this range the growth rate becomes an exponential Arrhenius function more typically observed in chemical vapor deposition. To ensure ALD-like growth conditions a substrate temperature of 200° C. was chosen for all of the samples considered in this study.

Physico-Chemical Characterisation of Al.sub.2O.sub.3:F Films

[0070] To confirm the presence of fluorine in our samples, secondary ion mass spectrometry was used to provide a depth profile of the fluorine distribution (FIG. 1a) in the 10% and 100% cycle doped films. The distribution of aluminium (Al.sup.+, mass/charge ratio of m/z 27) initially increases from the surface and stabilizes after is of sputtering time). The other traces (m/z of 19) show a convolution of F.sup.− and O.sup.18H.sup.− ion distributions. The undoped (0% cycles) film illustrates the background 19 amu distribution, which is predominantly the hydroxyl ion mass interference. For this reason it was not feasible to measure the F-distributions in lower percentage-cycle doped films. The 10% and 100% cycle doped films show an increase in fluorine uptake with doping cycle fraction. Ellipsometry was used to establish the deposition rate (A/cycle) of ALD Al.sub.2O.sub.3 as a function of doping cycles at a growth temperature of 200° C.

[0071] Ellipsometry was also used to measure the refractive index dispersion, as it is sensitive to small changes in the chemistry of the films. The effect of the F incorporation can be revealed by measurement of the Sellmeier coefficients, if the overall absorption coefficient k is small [A8]. For the alumina—based films studied here, k˜0 and the refractive index n of the dielectric is given by:

[00001]$\begin{array}{cc}{n}^2\left(\lambda \right)=1+\underset{j}{.Math.}.Math.\frac{s_j.Math.{\lambda }^2}{{\lambda }^2-{\lambda }_j^2}& \left(1\right)\end{array}$

[0072] Where S.sub.j is the oscillator strength of an optical transition at wavelength λ.sub.j and λ is the wavelength of the incident light. Using the single oscillator model, it is assumed that one oscillator dominates, in the wavelength range used (500 nm-800 nm) and a single term Sellmeier relationship can be used:

n.sup.2(λ)−1=S.sub.0λ.sub.0.sup.2/[1−(λ.sub.0/λ).sup.2]  (2)

[0073] S.sub.0 effectively represents the average oscillator strength and λ.sub.0 is the average oscillator position. A measurement of the average interband-oscillator energy E.sub.0 (eV) is obtained from E.sub.0=hc/eλ.sub.0, where c is the speed of light, h is Planck's constant and e is the electronic charge. FIG. 1b shows a plot of n.sup.2−1 versus 1/λ.sup.2 for films of undoped alumina and with doping cycle percentages of 10% and 100%. The gradient of the straight line for each sample is a measure of 1/S.sub.0 and the intercept reflects 1/S.sub.0λ.sub.0. Lai et al [A9] have previously correlated S.sub.0 and λ.sub.0 with the concentration of fluorine incorporated into thin Al.sub.2O.sub.3 films either by .sup.19F.sup.+ ion implantation, CF.sub.4 plasma treatment or a combination of both. In their study they observe S.sub.o increases with increasing fluorine dose rate (5×10.sup.13 cm.sup.−2 to 2×10.sup.14 cm.sup.−2) and that λ.sub.0 also increases. For the films deposited using the in situ ALD doping process reported here, the same general trend is observed i.e. that the refractive index n decreases and λ.sub.0 increases with increasing exposure to F-doping cycles. However we note that S.sub.o is relatively insensitive to the F dosage in comparison with implanted/diffused material.

[0074] Transmission electron micrographs were recorded from cross sections of MOS capacitors (FIGS. 1c and d) fabricated from the Al.sub.2O.sub.3:F films, to ensure the thickness of the dielectric and reveal any interactions between the oxide and GaN substrate or top contact metal.

[0075] These data show the controlled introduction of fluorine during the ALD film growth. The [F] incorporation is proportional (SIMS and ellipsometry) to the addition of NH.sub.4F to a fraction of the ALD.

Example 2—Preparation of Metal-Oxide-Semiconductor (MOS Capacitors) (University of Glasgow Data)

[0076] MOS capacitors were made by a process of additive; subtractive; and lithographic steps in a sequence suitable for the fabrication of ohmic and rectifying contacts to the semiconductor and dielectric layers.

[0077] The test results for the capacitors formed with different alumina dielectric barrier layers are shown in FIGS. 2a to 2d.

[0078] FIG. 2a shows capacitance-voltage curves, which show F addition shifts the threshold voltage and reduces the hysteresis caused by defects in the oxide.

[0079] FIG. 2b shows dC/dV (y-axis) versus V.sub.Bias (x-axis) and illustrates that F addition makes a positive threshold voltage.

[0080] The leakage current (I-V) data shown in FIG. 2c shows leakage suppression although 10% seems to reinstate leakage again—suggesting upper limit to benefit from doping

[0081] All samples benefit from forming gas annealing (FGA) at 430° C. for 30 minutes. FIG. 2d shows the effect of FGA on the performance of a 5% doping cycles, F-doped dielectric layer prepared according the methodology of the present invention.

Example 3—Metal-Oxide-Semiconductor Field Effect Transistors (MOSFET) (University of Sheffield Data)

[0082] MOSFET transistors were prepared by a process of additive; subtractive; and lithographic steps in a sequence suitable for the fabrication of ohmic and rectifying contacts to the semiconductor and dielectric layers.

[0083] To compare the effect of fluorine addition, one MOSFET was prepared with a 21.2 nm dielectric barrier layer of alumina doped with fluorine (200 cycles of 1% F-doped doping cycles, Al.sub.2O.sub.3+20 cycles of Al.sub.2O.sub.3) and a second MOSFET was prepared with a 22.2 nm alumina barrier layer (with no fluorine doping).

[0084] The results are shown in FIGS. 3a to 3f.

MOSFET Prepared with Dielectric Barrier Layer Formed from 200 Cycles of 1% Doping Cycles, F-Doped Al.sub.2O.sub.3+20 Cycles of Al.sub.2O.sub.3 (21.2 nm):
Maximum drain current between 400 and 700 mA/mm

V.SUB.TH.: ˜−6.5 V to −8 V

[0085] Peak gm˜70-100 mS/mm
Hysteresis observed with bi-directional sweeps
MOSFET Prepared with a 22.2 nm Barrier Layer Formed of Al.sub.2O.sub.3 (Non F-Doped):
Maximum drain current between 300 and 450 mA/mm

V.SUB.TH.: ˜−7 V to −8 V

[0086] Peak gm˜55-65 mS/mm
Hysteresis observed with bi-directional sweeps

[0087] In summary, fluorine addition in the transistors reduces hysteresis and causes a positive threshold voltage shift.

REFERENCES

[0088] A1. P. D. Ye, B. Yang, K. K. Ng, and J. Bude, “GaN metal-oxide semiconductor high-electron-mobility-transistor with atomic layer deposited Al.sub.2O.sub.3 as gate dielectric,” Appl. Phys. Lett., vol. 86, no. 9, p. 063 501, January 2005. [0089] A2. Y. Z. Yue, Y. Hao, Q. Feng, J. C. Zhang, X. H. Ma, and J. Y. Ni, “GaN MOS-HEMT using ultrathin Al.sub.2O.sub.3 dielectric grown by atomic layer deposition,” Chin. Phys. Lett., vol. 24, no. 8, pp. 2419-2422, August 2007. [0090] A3. Yuanzheng Yue, Yue Hao, Senior Member, IEEE, Jincheng Zhang, Jinyu Ni, Wei Mao, Qian Feng, and Linjie Liu, IEEE Electron Device Letters, vol. 29, NO. 8, AUGUST 2008 p. 838 [0091] A4. C. Liu, E. F. Chor, and L. S. Tan, “Investigations of HfO.sub.2/AlGaN/GaN metal-oxide-semiconductor high electron mobility transistors,” Appl. Phys. Lett., vol. 88, no. 9, p. 173 504, April 2006. [0092] A5. Y. Cai, Y. Zhou, K. M. Lau, and K. J. Chen, IEEE Trans. Electron Devices, vol. 53, no. 9, pp. 2207-2215, September 2006. [0093] A6. Chao Chen, Xingzhao Liu, Benlang Tian, Ping Shu, Yuanfu Chen, Wanli Zhang, Hongchuan Jiang, and Yanrong Li, IEEE Electron Device Letters, 32 (10), (2011) [0094] A7. Yuhao Zhang, Min Sun, Sameer J. Joglekar, Tatsuya Fujishima, and Tomas Palacios, Applied Physics Letters 103, O.sub.33524 (2013) [0095] A8. M. DrDomenico Jr. and S. H. Wemple, Journal of Applied Physics 40, 720 (1969) [0096] A9. Chao Sung Lai, Kung Ming Fan, Hsing Kan Peng, Shian Jyh Lin, Chung Yuan Lee and Chi Fong Ai, Applied Physics Letters 90, 172904 (2007) [0097] A10. R. H. French, D. J. Jones, and S. Loughin, J Amer. Ceram. Soc., 77(2) 412-422 (1994)