Results and Discussions
6.2 Electron Beam Lithography System
6.2.7 Comparison between CA Resist and PMMA/PMMA-MAA
Table 6-2 is a table comparing dosage/footprint between CA resist and PMMA/PMMA-MAA. Same footprint length can be achieved using these two different types of resists; however, the difference between dosages is significant.
Coulomb, C, is related to time and is defined by the following relation,
s A
C= • (8)
where A (ampere) is the current, and s (second) is the time. Because the beam current of the electron beam system is fixed, higher dosage means longer exposure time. This proves that using CA resist for T-shaped gate fabrication is more efficient than using PMMA. Take footprint length 170nm for example, if PMMA were used, the required dosage was 1600 µC/cm2. On the other hand, if CA resist were used, the required dosage was only 46µC/cm2. In this case, it means that the required exposure time of PMMA was 35×
longer than that of the CA resist.
Chapter 7
Conclusions
In this study, two novel methods for the submicron gate fabrication have been introduced. We had successfully demonstrated gate fabrication using an I-line stepper on a silicon substrate. Using the stage shifting mechanism, we were able to shrink the gate length to more than half the size of the original pattern. In this study, we had demonstrated γ-shaped gates with gate length equal to 0.2 and 0.3µm. In addition, we had shown that 89% of the errors are within 45nm and the average error is 23nm, which made this method of gate fabrication very reliable. Metal was also successfully deposited and lifted-off to form a T-gate.
Another method which had been introduced was to fabricate T-shaped gate with electron beam lithography system and the chemical amplified resist.
We had demonstrated that using chemical amplified photoresist to fabricate gate was more efficient than using PMMA/PMMA-MAA. The required exposure time of PMMA/PMMA-MAA was 35× longer than that of the chemical amplified resist. A MHEMT device with 0.55 of In content in the channel was also used to demonstrate this method. The gate length for this device was around 0.25µm and the saturation drain current was 360mA/mm.
Furthermore, the breakdown voltage was around 5.1V. Furthermore, the measured transconductance, gm, was 760mS/mm which was comparable to other MHEMT devices with the same gate length
In conclusion, the two novel methods proposed in this study both showed
fabricating submicron T-shaped gate for III-V semiconductor high frequency devices for wireless applications.
Substrate Buffer Layer Channel Layer Doping Region
Cap Layer Drain
Source Gate
Figure 2-1 Typical HEMT Device strucuture
Current Flow Recess region
Figure 2-2 Two-dimensional electron gas (2DEG) at AlGaAs/GaAs heterinterface
Electrons
AlGaAs
GaAs 2DEG
Mesa Isolation
Ohmic Contact
T-Shaped Gate
Gate Recess
Passivation + Contact Via Airbridge
Figure 3-1 Basic Flow Chart of HEMT Fabrication
Figure 3-2 Typical Air Bridge a) Top-View; b) Side-View Gate
Source
Drain
Air Bridge
a)
b)
Substrate PMMA-MAA
PMMA
E-beam exposure and development
Substrate PMMA-MAA
PMMA
Metal deposition and lift-off
Substrate
Figure 4-1 Dual layer gate fabrication process
Substrate
Figure 4-2 Modified Dual Layer Process
a) b)
d) c )
e) f)
Figure 4-3 Flowchart of Gate Fabrication Using I-line Stepper
Figure 4-4 Stepper Alignment Mark: a) Top View; b) Side View a)
b)
Figure 4-5 Location of Grating on a Stepper Mask Preset Point
Si Wafer Si3N4 Si3N4
Top View Side View
x
Figure 4-7 Stage shifted after alignment
Si Wafer Si3N4 Si3N4
Top View Side View
Figure 4-6 Mask aligned exactly over the first pattern
0.4 µm
Figure 4-8 I-line stepper mask used for this study
Soft Bake
E-beam exposure and development
Post exposure bake
Nitride etching using ICP PR spin coated
Hard Bake
PR spin coated
Soft Bake
E-beam exposure and development
Hard Bake
Gate metal deposition and lift-off
Figure 4-9 Flowchart of gate fabrication using chemical amplified resist
a1 = in at input
b1 = out at input
b2 = out at output
a2 = in at output
Figure 5-1 Definition of S-parameters S11 = b1/a1Input reflection coefficient
S21 = b2/a1Gain/Loss S12 = b1/a2Isolation
S22 = b2/a2Output reflection coefficient
Figure 6-1 Scattering Effect
Figure 6-2a) SEM picture after shifting mechanism
Figure 6-2b) SEM picture of a 0.2µm shifted sample
Figure 6-3 Optimum result of a 0.2µm shifted sample
Figure 6-4 SEM picture of a 0.1µm shifted sample
Figure 6-5 Metal gate after lift-off process
0 50 100 150 200 250 300
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Reticle
Final Gate Length (nm)
Figure 6-6 Bar graph of 18 reticles with 0.2µm shift
Slot
Sample
(a)
(b)
Figure 6-7 SEM Stage
Figure 6-8a Foot for T-shaped Gate
Figure 6-8b Head for T-shaped gate
0.0 0.3 0.6 0.9 1.2 1.5 0
100 200 300 400 500
Vgs=-1V Vgs=-0.8V Vgs=-0.6V
Vgs=-0.4 Vgs=-0.2 Vgs = 0V
Drain-source current I D(mA/mm)
Drain-source voltage VD(V) Figure 6-9 I-V curve
0 1 2 3 4 5 6 0.0
0.2 0.4 0.6 0.8 1.0
VB=5.1V Drain-gate current I DG(mA/mm)
Drain-gate voltage VDG (V)
Figure 6-10 VDG vs. IDG
0
I DS (mA/mm)Drain-source current Transconductancegm (mS/mm)
VDS = 1.5 V
gm,max = 760 mS/mm
Figure 6-11 Transconductance, gm
35
40 42
44 46
48 50
52 54
56 60
120 140 160 180 200 220
33 38 43 48 53 58
Dosage (µC/cm2)
Footprint Length (nm)
Figure 6-12 Dosage vs. Footprint Length for CA resist
500 600 700 800 900 1000 1100
220 240 260 280 300 320
Dosage (µC/cm2)
Tee-Top Length (nm)
Figure 6-13 Dosage vs. Tee-Top Length for CA Resist
150 170 190 210 230 250 270 290
1550 1600 1650 1700 1750 1800 1850 1900 1950 2000 2050
Dosage (µC/cm2)
Footprint Length (nm)
Figure 6-14 Dosage vs. Footprint Length for PMMA
Reticle Final Gate Length (nm)
Table 6-1 Actual gate lengths and experimental errors
Dosage (µµµµC/cm2) Footprint Length (nm)
CA resist (DSE1010) 46 171
PMMA/PMMA 1600 170
CA resist (DSE1010) 56 210
PMMA/PMMA 1700 215
Table 6-2 Comparison between CA Resist and PMMA/PMMA-MAA
References
[1] Iain Thayne, “Advanced III-V HEMTs”, III-Vs REVIEW THE ADVANCED SEMICONDUCTOR MAGAZINE VOL 16 - NO 8 - NOVEMBER 2003.
[2] C.Y. Chang, S.M. Sze, “ULSI Technology”, The McGraw-Hill Companies, Inc.
[3] Kazuhiko Honjo, “Microwave communication semiconductor circuit”, Hikkan Kogyo Shinbunsha, Tokyo
[4] Allan W. Scott, “Understanding Microwaves”, John Wiley & Sons, Inc.
[5] J. Michael Golio,”Microwave MESFETs and HEMTs, Artech House, Boston, London
[6] Michael Quirk, Julian Serda, “Semiconductor Manufacturing Technology”, Pearson Prentic Hall
[7] Ralph Williams, “Modern GaAs Processing Methods”, Artech House
[8] Peter Van Zant, “Microchip Fabrication”, McGraw-Hill International
[9] F. Benkhelifa, M.Chertouk, M. Walther, R. Losch, G. Weimann,
“Metamorphic HEMT 0.5 µm Low Cost High Performance Process on 4”
GaAs Substrates”, Fraunhofer Institue for Applied Solid State Physics (IAF), Tullastrasse 72, D-79108 Freiburg, Germany
[10] V. Van Driessche, J. Finders, A. Tritchkow, K. Ronse, L. Van den hove, P. Tzviatkov, “Feasibility of 250 nm gate patterning using I-line with OPC”, Microelectronic Engineering 41/42 (1998) 111-116
[11] Edward Yi CHANG, Yueh-Chin LIN, Guan-Ji CHEN, Huang-Ming LEE, Guo-Wei HUANG, D. BISWAS and Chun-Yen CHANG,
“Composite-Channel Metamorphic High Electron Mobility Transistor for Low-Noise and High-Linearity Applications”, Japanese Journal of Applied Physics Vol. 43, No. 7A, 2004, pp. L 871–L 872
[12] 陳世哲，柯富祥 “I-line 光學步進機對準符號之設計法則”, 國家還微 米元件實驗室