You are on page 1of 7

50 IEEE TRANSACTIONS ON SEMICONDUCTOR MANUFACTURING, VOL. 19, NO.

1, FEBRUARY 2006

Interconnect Capacitance Characterization Using


Charge-Injection-Induced Error-Free (CIEF)
Charge-Based Capacitance Measurement (CBCM)
Yao-Wen Chang, Hsin-Wen Chang, Tao-Cheng Lu, Ya-Chin King, Wenchi Ting, Yen-Hui Joseph Ku, and
Chih-Yuan Lu, Fellow, IEEE

Abstract—In this work, we describe a novel operation of rors are first discussed. Next, its efficient layout structure is
charge-injection-induced error-free charge-based capacitance demonstrated, making CIEF CBCM very suitable for indus-
measurement (CIEF CBCM) method. This method has the sim- trial applications. Large amounts of experimental results for
plest test structure among various CBCM methods by using only
one N/PMOS pair. CIEF CBCM has the advantage of being free backend process monitoring within a limited layout area can
from charge-injection-induced errors and of efficient layout area be achieved. In the fourth section, the example exploring the
usage. It is very suitable for industrial applications for large impact of dummy metal fills in CMP process on interconnect
amounts of accurate capacitance characterizations with a limited capacitances is demonstrated [7]. The final section includes
layout area. Besides, CIEF CBCM is also implemented for inves- in-depth discussion on the optimization of the chip performance
tigating the impact of floating dummy metal fills on interconnect
capacitance directly from silicon data. and minimization of the crosstalk by suitable dummy pattern
Index Terms—Capacitance measurement, charge-based ca- design.
pacitance measurement (CBCM), charge injection, interconnect
capacitance. II. OPERATION OF CIEF CBCM
The test structure of CIEF CBCM comprises an N/PMOS pair
I. INTRODUCTION
in a pseudoinverter configuration and they are driven by two
nonoverlapping signals and , as shown in Fig. 1. The
HARGE-BASED capacitance measurement (CBCM) was
C first proposed in 1996 to characterize on-chip interconnect
capacitance with sub-femto-farad resolution [1]. It receives
capacitance under test is placed between the output of
the inverter and one probe pad, to which a controlled pulse or a
fixed bias voltage can be directly applied. The capacitance can
more and more attentions because the intermetal capacitances
be extracted by the two-step operations of CIEF CBCM. At the
have become a significant factor in the design of fast chips.
first step, we force the probe pad to ground. At this step, not only
Accurate characterization and modeling of interconnect ca-
the capacitance under test, but also all the parasitic capacitances
pacitance are essential now for successful circuit design [2],
at the output node are charged and discharged between
[3]. Following the concept of CBCM, the single pattern driver
and ground repeatedly. The charging current through pMOS
(SPD) method [4] was proposed, which is free from errors
is expressed as
caused by mismatch issues and allows all the coupling capaci-
tance extractions. However, errors from charge injection always
exist and limit the resolution of CBCM methods [4]. To reduce (1)
the impact of charge injection, an improved CBCM scheme [5]
employing pass gates was proposed. The errors resulting from where is the frequency of the nonoverlapping clock signals.
charge injection are suppressed but still exist. Following that, an At the second step, we apply a controlled pulse to the probe
improved CBCM method, charge-injection-induced error-free pad as shown in Fig. 1. The pulse rises up from ground to
(CIEF) CBCM method was proposed [6]. It is completely free before pMOS is turned on. When pMOS is turned on ,
from the errors induced by charge injection and more suitable the output node is pulled up to . There will be no charges at
for the extractions of interconnect capacitances. both sides of because there is no voltage drop on it. Next,
In this paper, the operation of CIEF CBCM and its special before nMOS is turned on to discharge the output node to
advantage of freedom from the charge-injection-induced er- ground, the controlled pulse falls down to ground to keep no
charges at both sides of the capacitance. Thus, only the current
that charges up the parasitic capacitances is recorded as
Manuscript received July 19, 2005; revised October 10, 2005.
Y.-W. Chang is with the Macronix International Co., Ltd. no. 16, Hsinchu (2)
300, Taiwan, R.O.C and also with the Institute of Electronics Eng., National
Tsing-Hua University, Hsinchu 300, Taiwan, R.O.C. (e-mail: wenchang@mxic.
com.tw). Under such operations, can then be extracted by sub-
H.-W. Chang, T.-C. Lu, W. Ting, Y.-H. J. Ku, and C.-Y. Lu are with the tracting from as shown in (3)
Macronix International Co., Ltd. no. 16, Hsinchu 300, Taiwan, R.O.C.
Y.-C. King is with the Institute of Electronics Eng., National Tsing-Hua Uni-
versity, Hsinchu, Taiwan 300, R.O.C.
Digital Object Identifier 10.1109/TSM.2005.863228
(3)

0894-6507/$20.00 © 2006 IEEE


CHANG et al.: INTERCONNECT CAPACITANCE CHARACTERIZATION USING CIEF CBCM 51

Fig. 1. The proposed CIEF CBCM structure and the bias setup in two-step measurement. The dashed line shows the applied bias at the probe pad in the first step
and the solid line shows the applied bias in the second step.

Fig. 2. The simulated charge-injection-induced errors versus C for the


conventional CBCM, SPD, and CIEF CBCM.

In conventional CBCM operations, charge injection, which


always occurs at the transition when the MOSFET transistors
are turned off [8], will cause the inevitable errors and limits the
resolution of conventional CBCM methods. There will be un-
wanted flowing-back currents during the turnoff transitions and
the currents depend on the capacitance at the output node of
the pseudoinverter. In conventional CBCM and SPD, the ca-
pacitances are not the same at each operation step. Therefore,
the flowing-back currents are different, which then induces the
errors. We emulate the operations of conventional CBCM and
SPD methods in SPICE simulator and Fig. 2 shows the simu-
lated errors that result from charge injection. It shows that the
errors increase with the decreasing capacitance under test. But
in CIEF CBCM operation, the capacitance at the output node is
Fig. 3. (a) The layout arrangement of CIEF CBCM for single capacitance ex-
always the same at two steps. The only difference is the bias traction. In total, 20 capacitances can be extracted by sharing one pseudoinverter
applied to the pad is ground at the first step and at the in one 24-pad spine. (b) The layout arrangement of CIEF CBCM with multielec-
second step when pMOS switches off. It makes no difference on trode pattern. There are three intermetal capacitances can be extracted in one
multielectrode pattern. Eighteen sets of pseudoinverters driving the same mul-
charge injection behavior because they both are dc bias during tielectrode structures are put in one 24-pad spine and, in total, 54 capacitances
this phase. As a result, the charge-injection-induced currents can can be extracted.
be completely canceled out so that no errors will be generated.
Ideally, CIEF CBCM shows the perfect results and is with accuracy as small as picoampere under the measurement
completely free from the errors induced by charge injection. system, attofarad resolution can be achieved when CIEF CBCM
However, in practice, the limitation of the measurement system is operated under frequency of megahertz range. Besides, there
would limit the resolution of the method. If we can get current might be additional errors due to the small differences in the
52 IEEE TRANSACTIONS ON SEMICONDUCTOR MANUFACTURING, VOL. 19, NO. 1, FEBRUARY 2006

Fig. 4. Diagram of CIEF CBCM structure driving a multielectrode pattern. The pattern has three M1 lines sandwiched by M2 and poly plates. The central M1
line is driven by the pseudoinverter of CIEF CBCM, and the neighboring M1 lines, the top M2 plate, and the bottom poly plate are connected to the probe pads
respectively.

potentials on both sides of in the second step, which can


be eliminated by careful calibrations of the added pulse.

III. EFFICIENT LAYOUT STRUCTURE AND


EXPERIMENTAL RESULTS
CIEF CBCM, using only one pair of pseudoinverters, is the
simplest one among various CBCM methods. With the pseu-
doinverter shared, only one probe pad is needed for each ad-
ditional capacitance characterization as shown in Fig. 3(a). By
CIEF CBCM, we can extract 20 different interconnect capaci-
tances in one 24-pad spine. The conventional CBCM and SPD
both need additional two probe pads for each additional ca-
pacitance. CIEF CBCM occupies the least layout area and is
Fig. 5. The measured coupling capacitances of the central M1 lines to the
the most efficient method for interconnect characterization and neighboring M1 lines, to the top M2 plate, and to the bottom Poly plate, versus
process monitor. . overlapped length with fixed spacing of 0.32 m
the
The efficiency can be further enhanced when CIEF CBCM
is used to characterize the cross-coupling capacitances in mul-
tielectrode structure. That is, the structure has more than two
electrodes and there is more than one intermetal capacitance
that can be extracted. Fig. 4 shows our experimental multielec-
trode structure under CIEF CBCM connection fabricated by
three-metal-layer 0.13- m Flash technology. The structure has
three M1 lines sandwiched by M2 and poly plates. The width of
M1 lines is designed as 0.32 m and the spacing between M1
lines ranges from 0.32 to 5 m. Besides, the overlapped length
is changed from 5 to 50 m. The central M1 line is driven by the
pseudoinverter of CIEF CBCM, and the neighboring M1 lines,
the top M2 plate, and the bottom poly plate are connected to the
probe pads, respectively. Thus, three intermetal capacitances in Fig. 6. The measured coupling capacitances of the central M1 lines to the
the same structure by CIEF CBCM operations can be extracted. neighboring M1 lines, to the top M2 plate, and to the bottom Poly plate, versus
They are the coupling capacitances from the central M1 line to the spacing with fixed overlapped length of 50 m.
the neighboring M1 lines, to the top M2 plate, and to the bottom
poly plate respectively. Except three common pads for , Fig. 5 shows these three capacitances versus the overlapped
and ground, if we arrange multiple CIEF CBCM structures as length with fixed spacing of 0.32 m; and Fig. 6 shows these
shown in Fig. 3(b) by connecting all neighboring M1 lines to capacitances versus the spacing with fixed overlapped length of
pad E1, all M2 plates to pad E2, and all poly plates to pad E3, 50 m. Fig. 5 indicates that all these capacitances show good lin-
in total 18 sets of pseudoinverters driving the similar multielec- earity with the overlapped length and there are always positive
trode structures can be put in one 24-pad spine. A total of 54 y-intercepts, which should come from the fringing capacitors at
capacitances from only one spine can be extracted. It is very ef- edges of the parallel lines. In Fig. 6, the coupling capacitance
ficient and suitable for industrial applications to collect a large to the neighboring M1 lines decreases with increasing spacing.
amount of silicon data without occupying a large layout area. But the capacitances to the top M2 plate and to the bottom poly
CHANG et al.: INTERCONNECT CAPACITANCE CHARACTERIZATION USING CIEF CBCM 53

Fig. 7. Structure having three parallel M2 lines above one poly line
with dummy M1-fills filled. The width/length/spacing of the M2 lines are
5 m=150 m=1 m. The poly line is right below the central M2 with the
same size.
Fig. 9. SEM picture of the backend profile. This is the region with dummy-M1
fill. The thickness of M1 is 7600 
A; the thickness of IMD (including M1) is
14 500 A; and the thickness of M2 is 6500 A.

Fig. 10. Measured and simulated overlapping capacitances of the structures


with different dummy-fill patterns.
Fig. 8. Three different dummy-fill patterns with each block having dimen-
sion of 5 m by 10 m. (a) Basket-weave type. (b) Perpendicular alignment
with dummy-fills right under the central M2 line. (c) Perpendicular alignment of dummy fill is 5 m by 10 m. Pattern A is the basket-weave
without dummy-fills under the central M2 line. type, which is the suggested pattern to minimize the variations
of the capacitances coming from floating dummy-fills [10]
plate increase inversely with increasing spacing. All these ca- and is the default pattern in this process. Pattern B and C are
pacitances are extracted from the test patterns in one spine only. both the cases of perpendicular alignment. In pattern B, one
line of M1-fills are directly under the middle M2 line; but in
IV. INVESTIGATION INTO THE IMPACT OF FLOATING pattern C, oppositely, there are no M1-fills under the middle
DUMMY-FILLS ON INTERCONNECT CAPACITANCE M2 line. In addition, the structure without any dummy M1-fill
With the help of CIEF CBCM, we can characterize the is also added and compared. The thickness of intermetal di-
interconnect capacitances directly from silicon when increas- electric (IMD) between M1 and M2 layers can be gotten from
ingly complex backend process is adapted, which could not be a scanning electron microscope (SEM) picture (Fig. 9) for
implemented easily in the past. Next, the impact of floating simulation purpose. The thickness of IMD layer is the same
dummy-fills on interconnect capacitances is investigated using with or without dummy-fills because it is only a small region
CIEF CBCM. In the chemical–mechanical polishing (CMP) to block the filling of dummy-fills. Fig. 10 shows the extracted
process, in order to control the uniformity of interlayer di- overlapping capacitances of the central M2 line to the poly line
electrics, the dummy-fills have always been used [9], [10] and Fig. 11 shows coupling capacitances to the neighboring M2
and they would affect the interconnect capacitances more or lines. Also shown in the figures are the simulated interconnect
less [10]–[12]. It may result in a speed performance different capacitances by Raphael [13]. The simulation results correlate
from designer’s expectation. Fig. 7 shows our experimental to the measurement well and show the same trends among
structures with three parallel M2 lines above one poly line, structures with different dummy-fill patterns.
which are fabricated by the same three-metal-layer 0.13- m Fig. 12 shows the capacitance differences of coupling capaci-
Flash technology. All these four conductor lines are 5- m wide tance, overlapping capacitance, and the total capacitance, which
and 150- m long. The spacing between M2 lines is 1 m. is the sum of the former two components. The capacitances in
Among these conductors, three different patterns of floating the structures with pattern B, pattern C dummy-fills, and without
dummy M1-fills are filled as shown in Fig. 8(a)–(c). Each block dummy-fills are compared to those in the structure with pattern
54 IEEE TRANSACTIONS ON SEMICONDUCTOR MANUFACTURING, VOL. 19, NO. 1, FEBRUARY 2006

Fig. 11. Measured and simulated coupling capacitances of the structures with
Fig. 13. Simulated capacitance changes of: (a) coupling capacitances and
different dummy-fill patterns.
(b) total capacitances in the structure without dummy-fills, compared to those
in the structure with pattern A dummy-fills. Simulations are done using the
structures with different M2 widths from 1 to 5 m, and spacing from 0.5 to
2.5 m.

Fig. 12. Capacitance changes of the structures with pattern B, C, and no


dummy-fills, compared to that with pattern A.

A dummy-fills. We can find that when there are no dummy-fills,


the total capacitance and coupling capacitance are lower by 10%
and 20%, respectively. In other words, if the impact of floating Fig. 14. Simulated capacitance changes of: (a) coupling capacitances and (b)
dummy-fills is not considered in the circuit design, designers total capacitances in the structure with pattern B dummy-fills, compared to those
in the structure with pattern A dummy-fills. Simulations are done under the
may seriously underestimate the delay and crosstalk. When pat- structures with different M2 widths from 1 to 5 m, and spacing from 0.5 to
tern B dummy-fills are used instead of pattern A, the overlap- 2.5 m.
ping capacitance increases by about 28% because the dummy
M1-fills are right below the central M2 line and enhance the ca- be helpful if there is a signal line very sensitive to the crosstalk
pacitance increment. At the same time, the coupling capacitance from neighboring conductor line. The cost is only slight total
drops off more than 10% is observed. capacitance increment as shown in Fig. 14(b). Nevertheless,
when the M2 lines are as narrow as 1 m, the neighboring
V. DISCUSSION M2 lines are almost right above the dummy-fills. The floating
The above characterization results suggest that the floating dummy-fills would redirect some of the electric force lines
dummy-fills indeed play an important role on interconnect back to the neighboring M2 lines rather than direct almost all
capacitance. And it seems that the improvement of the circuit the electric force lines to the bottom poly line. As a result, the
performance can be expected by choosing suitable dummy-fill overlapping capacitances do not increase as expected and the
pattern. Through simulations of the same structures with dif- coupling capacitance reductions are not so significant.
ferent M2 widths from 1 to 5 m, and spacing from 0.5 to
2.5 m, we can demonstrate the effects in more details. Com- VI. CONCLUSION
pared to the structure with pattern A, both total capacitances In this work, we describe the operation of CIEF CBCM
and coupling capacitances can be effectively reduced by method and its special advantage of freedom from the errors
blocking the filling of the dummy-fills as shown in Fig. 13(a) induced by charge injection. Also, its superb efficiency for
and (b). And the capacitance reductions enlarge with increasing large amounts of capacitance characterizations within a limited
M2 width and spacing. Thus, if blocking the insertion of layout area is demonstrated. Both benefits make CIEF CBCM
dummy-fills within a limited region is allowed, it is the most the best solution for industrial applications to characterize
effective way to reduce the interconnect capacitances and the interconnect capacitances under different environments or
enhance the circuit speed. If blocking the filling of dummy-fills to monitor the variations of backend process. Also, with the
is not allowed, pattern B dummy-fills may be also effective to help of CIEF CBCM, a complete investigation of the impact
reduce the coupling capacitances as shown in Fig. 14(a). It may of the floating dummy-fills on interconnect capacitances is
CHANG et al.: INTERCONNECT CAPACITANCE CHARACTERIZATION USING CIEF CBCM 55

performed. It is confirmed directly from silicon that the impact Hsin-Wen Chang was born in Taichung, Taiwan,
is very significant and it could play an important role for R.O.C., in 1976. She received the B.S. degree in
physics from National Normal University, Taipei,
successful circuit designs. To optimize the chip performance Taiwan, R.O.C., in 1998, and the M.S. degree
and minimize the crosstalk, dummy pattern design should be in physics from National Tsing-Hua University,
carefully considered combining the circuit functionality and Hsinchu, Taiwan, R.O.C., in 2002.
In 2002, she joined Macronix International Co.,
interconnection routing environment. Ltd. (MXIC), Hsinchu, Taiwan, R.O.C., and has been
with the Device Engineering Department, where she
has engaged in MOS device characterization and
REFERENCES modeling, interconnect capacitance characterization,
and high-voltage device modeling.
[1] J. C. Chen, B. W. McGaughy, D. Sylvester, and C. Hu, “An on-chip, at-
tofarad interconnect charge-based capacitance measurement (CBCM)
technique,” in Proc. IEDM 1996, pp. 3.4.1–3.4.4.
[2] X. W. Lin and D. Pramanik, “Future interconnect technologies and
copper metallization,” Solid State Technol., vol. 41, pp. 63–79, Oct.
1998. Tao-Cheng Lu was born in Taiwan in 1967. He
[3] D. Sylvester and C. Hu, “Analytical modeling and characterization received the B.S., M.S., and Ph.D. degrees in elec-
of deep-submicrometer interconnect,” Proc. IEEE, vol. 89, no. 5, pp. trical engineering from National Taiwan University,
634–664, May 2001. Taipei, Taiwan, R.O.C., in 1989, 1991, and 1993
[4] B. Froment, F. Paillardet, M. Bely, J. Cluzel, E. Granger, M. Haond, respectively.
and L. Dugoujon, “Ultra low capacitance measurements in multilevel He joined the Macronix International Co., Ltd. in
metallization CMOS by using a built-in electron-meter,” in Proc. IEDM 1995, Hsinchu, Taiwan, R.O.C., where he engaged
1999, pp. 37.2.1–37.2.4. in the simulation and modeling of CMOS and
[5] T. Kunikiyo, T. Watanabe, T. Kanamoto, H. Asazato, M. Shirota, K. nonvolatile memory devices. From 1997 to 1999,
Eikyu, Y. Ajioka, H. Makino, K. Ishikawa, S. Iwade, and Y. Inoue, he set up the electrostatic discharge (ESD) group
“Test structure measuring inter- and intralayer coupling capacitance of in the device department and became the manager
interconnection with subfemtofarad resolution,” IEEE Trans. Electron of the device department in 2000. Currently, he is the Deputy Director of
Devices, vol. 51, no. 5, pp. 726–735, May 2004. the Device Technology Division and is in charge of process/device simula-
[6] Y. W. Chang, H. W. Chang, C. H. Hsieh, H. C. Lai, T. C. Lu, W. tion, device modeling, ESD/latchup, and nonvolatile memory device design,
Ting, J. Ku, and C. Y. Lu, “A novel simple CBCM method free from including the development of nitride trapping storage flash devices. He has
charge injection-induced errors,” Electron Device Lett., vol. 25, no. 5, published more than 40 papers in technical journals and conferences and has
pp. 262–264, May 2004. authored and coauthored more than 70 U.S. patents. His research interests
[7] Y. W. Chang, H. W. Chang, T. C. Lu, Y. King, W. Ting, J. Ku, and C. Y. include device/flash cell reliability and scaling, device modeling, interconnect
Lu, “A novel CBCM method free from charge injection induced errors: capacitance measurement and modeling, and on-chip ESD protection.
investigation into the impact of floating dummy-fills on interconnect
capacitance,” in Proc. ICMTS 2005, pp. 235–238.
[8] B. J. Sheu and C. Hu, “Switch-induced error voltage on a switched
capacitor,” IEEE J. Solid-State Circuits, vol. SC-19, no. 4, pp. 519–525, Ya-Chin King was born in Taiwan. She received the
Aug. 1984. B.S. degree in electrical engineering from National
[9] D. O. Ouma, D. S. Boning, J. E. Chung, W. G. Easter, V. Saxena, Taiwan University, Taipei, Taiwan, R.O.C., in 1992,
S. Misra, and A. Crevasse, “Characterization and modeling of oxide and the M.S. degree in electrical engineering and
chemical-mechanical polishing using planarization length and pattern the Ph.D. degree in thin oxide technology and novel
density concepts,” IEEE Trans. Semicond. Manuf., vol. 15, no. 2, pp. quasi-nonvolatile memory from the University of
232–244, May 2002. California, Berkeley, in 1994 and 1999, respectively.
[10] A. B. Kahng, G. Robins, A. Singh, and A. Zelikovsky, “Filling al- She joined the faculty of National Tsing-Hua Uni-
gorithms and analyses for layout density control,” IEEE Trans. Com- versity (NTHU), Hsinchu, Taiwan, R.O.C., in 1999.
puter-Aided Design Integr. Circuits Syst., vol. 18, no. 4, pp. 445–462, She is currently an Associate Professor in the Elec-
Apr. 1999. trical Engineering Department, NTHU. Her research
[11] B. E. Stine, D. S. Boning, J. E. Chung, L. Camilletti, F. Kruppa, E. R. topics include advance gate dielectric, CMOS image sensor, and nonvolatile
Equi, W. Loh, S. Prasad, M. Muthukrishnan, D. Towery, M. Berman, memory design.
and A. Kapoor, “The physical and electrical effects of metal-fill pat-
terning practices for oxide chemical-mechanical polishing processes,”
IEEE Trans. Electron Devices, vol. 45, no. 3, pp. 665–679, Mar. 1998.
[12] W. Lee, K. Lee, J. Park, T. Kim, Y. Park, and J. Kong, “Investigation of
Wenchi Ting received the Ph.D. degree in electrical engineering from the Uni-
the capacitance deviation due to metal-fills and the effective intercon-
versity of Texas, Austin.
nect geometry modeling,” in Proc. Symp. Quality Electronic Design,
He is currently a Senior Director at Macronix International Company,
2003, pp. 373–376.
Hsinchu, Taiwan, R.O.C., focusing on developing silicon-nitride trapping
[13] Raphael Reference Manual, Avant! Corp..
based nonvolatile memory processes. His research interests include developing
SRAM, EPROM, EEPROM, Flash, and embedded Flash technologies.

Yao-Wen Chang was born in Kaohsiung, Taiwan,


R.O.C., in 1972. He received the B.S and M.S. de-
grees in electrical engineering from National Taiwan Yen-Hui Joseph Ku received the B.S. degree in
University, Taipei, Taiwan, R.O.C., in 1994 and electrical engineering from National Cheng-Kung
1996, respectively. He is currently working toward University, Tainan, Taiwan, R.O.C., in 1979, and the
the Ph.D. degree at National Tsing-Hua University, M.S. and Ph.D. degrees in electrical and computer
Hsinchu, Taiwan, R.O.C. engineering from the University of Texas, Austin, in
In 1998, he joined Macronix International Co. Ltd. 1983 and 1988, respectively. His research was on the
(MXIC), Hsinchu, and has been with the Device En- self-aligned silicide and shallow junction formation
gineering Department, where he has engaged in MOS for deep submicrometer device application.
device characterization and modeling, interconnect In 1987, he joined Rapro Technology, Fremont,
capacitance characterization, high-voltage device modeling, and RF CMOS de- CA, as a cofounder with the responsibility on
vice modeling. RTPCVD reactor design and process development.
56 IEEE TRANSACTIONS ON SEMICONDUCTOR MANUFACTURING, VOL. 19, NO. 1, FEBRUARY 2006

From 1991 to 1992, he was with Paradigm Technology, San Jose, CA, where he nology with high-density DRAM/SRAM. In 1994, he became the cofounder
worked on 0.35-m 4-Mb SRAM technology development. In 1992, he joined of Vanguard International Semiconductor Corporation, which is a spin-off
LSI Logic, Santa Clara, CA, where he worked on the advanced logic technology memory IC company from ITRI’s Submicron Project. He was the Vice Presi-
development and served as integration manager of the R&D Division until dent of Operations, Vice President of R&D, and later President from 1994 to
1997. He later joined the Technology Development Division of the MXIC, 1999. He is now the Chairman and CEO of Ardentec Corp., Hsinchu, Taiwan,
Hsinchu, Taiwan, R.O.C. He currently leads the Technology Development R.O.C., a very large scale integration (VLSI) testing service company; and also
Center and is responsible for the development of advanced nonvolatile memory serves Macronix International, Hsinchu, as a Senior Vice President/CTO. He
and embedded SOC technologies. He has published more than 40 papers in led MXIC’s technology development team to successfully achieve the state
technical journals and conferences. of the art nonvolatile memory technology and now responsible for MXIC’s
overall memory operation. He has published more than 100 papers and has
been granted 123 international patents.
Dr. Lu is a Fellow of the American Physical Society. He was granted the
Chih-Yuan Lu (M’78–SM’84–F’95) received the highest honor prize—the National Science and Technology Achievement Award
B.S. degree from National Taiwan University, Taipei, by the Prime Minister of Taiwan, due to his leadership and achievement in
Taiwan, R.O.C., in 1972 and the Ph.D. degree in the Submicron Project. He also received the IEEE Millennium Medal and the
physics from Columbia University, New York, in most prestigious semiconductor R&D award in Taiwan from Pan Wen Yuan
1977. Foundation.
He was a Professor at National Chiao-Tung
University and with AT&T Bell Labs from 1984
to 1989. He later joined ERSO/ITRI in 1989 as a
Deputy General Director responsible for the MOEA
grand Submicron Project. This project successfully
developed Taiwan first 8-in manufacturing tech-

You might also like