The Stacked Capacitor DRAM Cell and
`Three-Dimensional Memory
`
`Mitsumasa Koyanagi, Department of Bioengineering and Robotics, Tohoku University, Japan
`
`koyanagi@sd.mech.tohoku.ac.jp
`
`TECHNICAL LITERATURE
`
`Device scaling and the DRAM cell
`The key component in a computer system is memo-
`ry, since both data and instructions in a stored-pro-
`gram-type computer are stored in main memory. Mag-
`netic core memories, used in the early stages of com-
`puters, were replaced by semiconductor memories
`early in the 1970’s. The first high density semicon-
`ductor memory was 1-Kbit dynamic random access
`memory (DRAM), developed by Intel. Since the
`advent of this 1-Kbit DRAM, the packing density and
`capacity of DRAM have continued to increase to
`today’s 4Gbit. Such increases were achieved by the
`evolution of the memory cell from a four-transistor-
`type, three-transistor-type to a one-transistor- type.
`The invention of the one-transistor-type cell (1-T cell)
`by Robert Dennard especially accelerated the evolu-
`tion of DRAM in conjunction with his device scaling
`theory [1, 2].
`
`Scaling limitation of the planar (2D) DRAM cell
`The one-transistor-type cell (1-T cell) consists of one
`transistor and one capacitor. A transistor acts as a
`switch, and the signal charges are stored in a capac-
`itor. The first 1-T cell was realized using one switch-
`ing transistor and one MOS capacitor. The number of
`signal charges stored in the storage capacitor has to
`be maintained at almost a constant, or can be only
`slightly reduced, as the memory cell size is scaled
`down. However, MOS capacitor value -- and hence
`the amount of signal charges – is significantly
`reduced as the memory cell size is reduced, even if
`the capacitor oxide thickness is scaled-down. There-
`fore, I forecast in 1975 that the 1-T cell with a two-
`dimensional (2D) structure using a planar MOS
`capacitor eventually would encounter a scaling-
`down limitation because we cannot reduce the MOS
`capacitor area according to scaling theory. In addi-
`tion, I pointed out that the use of a MOS capacitor
`in the 1-T cell would be a problem because the sig-
`nal charges are seriously reduced due to the influ-
`ence of the minority carriers generated in a silicon
`substrate. An inversion layer capacitance and a
`depletion layer capacitance are connected with the
`gate oxide capacitance in parallel in the MOS capac-
`itor. The charges in the inversion layer and the
`depletion layer are easily affected by the minority
`carriers, which are thermally or optically generated or
`generated by the irradiation of energetic particles in
`a silicon substrate. Therefore, I predicted that the 1-
`T cell using an MOS capacitor would encounter a
`scaling-down limitation due to the influence of the
`minority carriers as well.
`
`Invention of the three-dimensional (3D)
`DRAM cell
`In my Ph.D. research during 1971-1974 [3], I had
`commented on the silicon surface and the inversion
`layer in MOS structures. To evaluate the electrical
`properties of the interface states and the inversion
`layer, I myself built an impedance analyzer with the
`frequency range of 0.01Hz to 100MHz. I examined
`various kinds of capacitors, including high-k (high
`dielectric constant) capacitors as a reference capacitor
`of this impedance analyzer. Eventually, I made a vac-
`uum capacitor for a reference capacitor in which fin-
`type capacitor electrodes were encapsulated in a vac-
`uum container. From these studies, I learned that an
`ideal capacitor with low loss should consist of metal
`electrodes and a low loss insulator (MIM structure); a
`three-dimensional structure of capacitor electrodes is
`effective to increase the capacitance value, and there
`is a trade-off between high-k and loss in the capaci-
`tor insulator. In addition, I knew through my Ph.D.
`research that the charges in the inversion layer and
`the depletion layer are easily influenced by the minor-
`ity carriers. Therefore, I questioned why the MOS
`capacitor with inversion capacitance and the deple-
`tion capacitance was used as the storage capacitor in
`the 1-T cell when I first knew about it in 1975. Then,
`I tried to eliminate the inversion capacitance and the
`depletion capacitance by employing a passive capac-
`itor such as the MIM as a storage capacitor and thus
`proposed a three-dimensional (3D) cell in 1976 [4, 5].
`I called this new 3D memory cell a stacked capacitor
`cell (STC).
`
`Fabrication and evaluation of the three-dimen-
`sional stacked capacitor cell
`Figure 1 shows the basic structure of a stacked capac-
`itor cell (STC) where the storage capacitor is three-
`dimensionally stacked on a switching transistor [6, 7].
`A passive capacitor with the structure of an electrode-
`
`Fig. 1. Basic structure of stacked capacitor cell (STC).
`
`Winter 2008
`
`© 2008 IEEE
`
`IEEE SSCS NEWS
`
`37
`
`Page 1 of 5
`
`SAMSUNG EXHIBIT 2007
`NVIDIA v. SAMSUNG
`Trial IPR2015-01327
`
`

`
`TECHNICAL LITERATURE
`
`insulator-electrode is used as a storage capacitor. The
`bottom electrode of the storage capacitor is connect-
`ed to the source/drain region. I proposed to use self-
`aligned contacts to connect the bottom electrode of
`the storage capacitor and a bit line to the source/
`drain of the switching transistor. This self-aligned
`technique was also used for the formation of capac-
`itor electrodes. By three-dimensionally stacking the
`storage capacitor on the switching transistor we can
`dramatically reduce the memory cell area. In addi-
`tion, we can use a high-k material as a capacitor
`insulator to increase the storage capacitance, since a
`passive capacitor is used as a storage capacitor. This
`is also useful for reducing memory cell size. Fur-
`thermore, we can solve the problem that the signal
`charges in the inversion layer and depletion layer are
`influenced by the minority carriers since an inver-
`sion capacitance is not used in a stacked capacitor
`cell. In 1977, I fabricated the first DRAM test chip
`with a stacked capacitor cell using 3μ m NMOS tech-
`nology and presented a paper on the stacked capac-
`itor cell in 1978 IEDM (IEEE International Electron
`Devices Meeting) [6]. Figure 2 shows the SEM cross
`section of a stacked capacitor cell fabricated using
`3μ m NMOS technology.
`
`Fig. 3. Leakage-current versus applied-voltage character-
`istics for Si3N4 films with thin oxides on their surfaces.
`
`I also used a Ta2O5 film as a capacitor insulator for the
`first time [7]. Ta2O5 has a dielectric constant five or six
`times larger than that of SiO2. Therefore, we can
`greatly increase the storage capacitance, although the
`leakage current is larger compared to those of SiO2
`and Si3N4, as shown in Fig.4.
`
`Fig. 2. SEM cross section of a stacked capacitor cell fabri-
`cated using 3um nMOS technology.
`
`In this figure it is clearly shown that a storage capac-
`itor is stacked on a switching transistor and a self-
`aligned contact is successfully formed, although plas-
`ma etching and RIE (reactive ion etching) were not
`available at the time. The self-aligned contact is wide-
`ly used in today’s memory LSI’s. In this stacked
`capacitor cell, I employed polycrystalline silicon
`(poly-Si)- Si3N4 - polycrystalline silicon (poly-Si) as a
`storage capacitor. Thermal SiO2 had been used as a
`capacitor insulator in a conventional 1-T cell with a
`MOS storage capacitor. In the stacked capacitor cell,
`I used Si3N4 instead of SiO2 as a capacitor insulator
`to increase the storage capacitance. The dielectric
`constant of Si3N4 is approximately two times larger
`than that of SiO2. I found that the leakage current of
`Si3N4 was significantly reduced by oxidizing its sur-
`face, as shown in Fig.3.
`
`Fig. 4. Leakage current – applied electric field character-
`istics of storage capacitors. (a) poly-Si–SiO2–poly-Si, (b)
`poly-Si–SiO2 /Si3N4 (ON)–poly-Si, (c) poly-
`Si/Ta–Ta2O5–poly-Si.
`
`38
`
`IEEE SSCS NEWS
`
`Winter 2008
`
`Page 2 of 5
`
`

`
`In 1978, when I presented the first stacked capacitor
`cell paper in IEDM, it was revealed that the data
`retention characteristics of DRAM are seriously
`degraded due to “soft-error,” which is caused by the
`carriers generated by alpha-particle irradiation in the
`silicon substrate. At that time, I believed that a stacked
`capacitor cell is tolerant of soft-error, since the signal
`charges stored in the passive capacitor are not influ-
`enced by the carriers generated in the substrate. Fig-
`ure 5 shows the dependence of soft-error rate on
`cycle time in a DRAM test chip [8]. As I expected, the
`soft-error rate was dramatically reduced by employing
`a stacked capacitor cell.
`
`TECHNICAL LITERATURE
`A storage capacitor is stacked on the switching tran-
`sistor and the bit line in a top-capacitor-type cell, on
`the switching transistor in the intermediate-capacitor-
`type cell (original stacked capacitor cell) and on the
`isolation oxide (LOCOS) in the bottom-capacitor-type
`cell. The top-capacitor-type STC cell is another name
`for the capacitor-over-bit line (COB) stacked capacitor
`cell and widely used in current DRAM [10]. We can
`use various kinds of materials for the capacitor insu-
`lator and electrodes, and can employ low temperature
`processes in the formation of the storage capacitor in
`the COB-type stacked capacitor cell since the storage
`capacitor is formed on the top of the memory cell.
`
`Evolution of the three-dimensional (3D) memory
`cell and future memory
`In 1979, I fabricated 16K-bit DRAM using a stacked
`capacitor cell with the oxidized Si3N4 (O/N) capacitor
`insulator [9]. Then I tried to introduce a stacked
`capacitor cell in 64K-bit DRAM production. However,
`it was too early to do this due to cost. As a result, the
`oxidized Si3N4
`(O/N) capacitor insulator was
`employed in 64K-bit DRAM production. Since then,
`the oxidized Si3N4 (O/N) capacitor insulator has been
`widely used for DRAM production. A stacked capaci-
`tor cell was employed in a 1Mbit DRAM production
`for the first time by Fujitsu [11]. Hitachi also employed
`a stacked capacitor cell in 4Mbit DRAM production
`[12]. Many other DRAM companies used a trench
`capacitor cell in the early stage of 4Mbit DRAM pro-
`duction. However, the stacked capacitor cell, which
`eventually came to occupy a major position in 4Mbit
`to 4Gbit DRAM’s, has evolved by introducing the
`three-dimensional capacitor structures with fin-type
`electrode [13] and cylindrical electrode [14] in con-
`junction with a capacitor electrode surface morpholo-
`gy of hemi-spherical grain (HSG) [15]. In addition to
`the introduction of the three-dimensional capacitor
`structure, a storage capacitor insulator with high
`dielectric constant (high-k) was employed in high
`density DRAM’s. In general, the leakage current of
`high-k material increases by high temperature pro-
`cessing. Therefore, a COB-type stacked capacitor cell
`is suitable for introducing a high-k capacitor insulator
`since the storage capacitor can be formed by a lower
`temperature process. Thus, the Ta2O5 capacitor insu-
`lator was employed in 64Mbit and 256Mbit DRAM’s
`with a COB-type stacked capacitor cell. Since then,
`various kinds of high-k materials have been studied as
`storage capacitor insulators in high density DRAM’s
`beyond 1Gbit. The concept of the COB-type stacked
`capacitor cell -- that various kinds of materials can be
`stacked on the switching transistor using a lower tem-
`perature process -- has been carried on in new mem-
`ories with a three-dimensional structure such as Fe-
`RAM (Ferroelectric RAM), P-RAM (Phase Change
`RAM), R-RAM (Resistive RAM) and M-RAM (Magnetic
`RAM).
`
`Fig. 5. Dependence of soft-error rate on cycle time in 16
`K-bit DRAM test chip.
`
`Eventually I proposed three types of stacked capaci-
`tor cells as shown in Fig.6 [9].
`
`Fig. 6. Cross-sectional configuration of three types of
`stacked capacitor cells. (a) top-capacitor-type cell, (b) inter-
`mediate-capacitor-type cell, (c) bottom-capacitor-type cell.
`
`In a high density stacked capacitor DRAM beyond 16
`Gbit, a twitching transistor with a three-dimensional
`structure such as a Fin-FET and a vertical transistor
`
`Winter 2008
`
`IEEE SSCS NEWS
`
`39
`
`Page 3 of 5
`
`

`
`TECHNICAL LITERATURE
`
`Fig. 7. SEM cross section of 3D-DRAM test chip with ten memory layers fabricated by wafer-on-wafer bonding technol-
`ogy with through-silicon-vias (TSV’s).
`
`will be employed together with a cylindrical capacitor
`and high-k capacitor insulator. Furthermore, many
`DRAM chips eventually will be vertically stacked to
`realize 3D-DRAM in which the memory capacity dra-
`matically increases. We have already succeeded in
`fabricating a 3D-DRAM test chip with ten memory lay-
`ers as shown in Fig.7 [16, 17].
`
`This 3D-DRAM test chip was fabricated using a newly
`developed wafer-on-wafer bonding technology with
`through-silicon-vias (TSV’s) [18-20]. Such a 3D-DRAM
`can be directly stacked on a microprocessor chip to
`realize a 3D-microprocessor and to solve the prob-
`lems of memory data-bandwidth between the memo-
`ry and the processor. We also fabricated a 3D-micro-
`processor test chip in which a DRAM chip is stacked
`on a processor chip, as shown in Fig.8 [21].
`
`In the future, various kinds of LSI chips, sensor chips
`and MEMs chips will be vertically stacked into an
`ultimate 3D-LSI which we call a super-chip [17, 22].
`We have developed a new super-chip integration
`technology using a novel self-assembly method.
`
`References
`
`[3]
`
`[1]
`[2]
`
`R.H. Dennard, USP-3387286, June 4, 1968.
`R.H. Dennard, F.H. Gaensslen, H.N. Yu, V.L. Rideout,
`E. Bassous, and A.R. LeBlanc, “Design of Ion Implant-
`ed MOSFET’s with Very Small Physical Dimensions”,
`IEEE J. Solid-State Circuits, Vol. SC-9, pp. 256-268, Oct.
`1974.
`J. Nishizawa, M. Koyanagi and M. Kimura, “Imped-
`ance Measurements to Study Semiconductor Surface
`and Thin Insulating Films on Its Surface”, Japanese J.
`Applied Phys., Suppl. 2, pp.773-779, 1974.
`[4] M. Koyanagi and K. Sato, Japanese patent application,
`Tokugansho 51-78967, 1976.
`[5] M. Koyanagi and K. Sato, USP-4151607, Jan. 5, 1977.
`[6] M. Koyanagi, H. Sunami, N. Hashimoto and M.
`Ashikawa, “Novel High Density, Staked Capacitor
`MOS RAM”, IEDM Tech. Dig., pp. 348-351, Dec. 1978.
`[7] M. Koyanagi, H. Sunami, N. Hashimoto and M.
`Ashikawa, “Novel High Density, Staked Capacitor
`MOS RAM”, Extended Abstracts of the 10th Confer-
`
`Fig. 8 SEM cross section of a 3D-microprocessor test chip fabricated by wafer-on-wafer bonding technology with
`through-silicon-vias (TSV’s).
`
`40
`
`IEEE SSCS NEWS
`
`Winter 2008
`
`Page 4 of 5
`
`

`
`TECHNICAL LITERATURE
`
`ence on Solid State Devices, 1978.
`[8] M. Koyanagi, Y. Sakai, M. Ishihara, M. Tazunoki and
`N. Hashimoto, “16-kbit Stacked Capacitor MOS RAM”,
`IEICE Technical Report (Japanese), SSD80-30, 1980.
`[9] M. Koyanagi, Y. Sakai, M. Ishihara, M. Tazunoki and
`N. Hashimoto, “A 5-V Only 16-kbit Stacked Capacitor
`MOS RAM”, IEEE Trans. on Electron Devices, ED-8,
`pp.1596-1601, 1980/ IEEE J. Solid-State Circuits, SC-15,
`pp.661-666, 1980.
`[10] S. Kimura, Y. Kawamoto, T. Kure, N. Hasegawa, J.
`Etoh, M. Aoki, E. Takeda, H. Sunami and K. Itoh, “A
`New Stacked Capacitor DRAM Cell Characterized by a
`Storage Capacitor on a Bit-line Structure”, IEDM Tech.
`Dig., pp. 596-599, Dec. 1988.
`[11] Y. Takemae, T. Ema, M. Nakano, F. Baba, T. Yabu, K.
`Miyasaka, K. Shirai, “A 1Mb DRAM with 3-Dimension-
`al Stacked Capacitor Cells”, ISSCC Dig. Tech. Papers,
`pp. 250-251, Feb. 1985.
`[12] K. Kimura, K. Shimohigashi, J. Etoh, M. Ishihara, K.
`Miyazawa, S. Shimizu, Y. Sakai, and K. Yagi, “A 65-ns
`4-Mbit CMOS DRAM with a Twisted Driveline Sense
`Amplifier”, IEEE J. Solid-State Circuits, SC-22, pp.651-
`656, 1987.
`[13] T. Ema, S. Kawanago, T. Nishi, S. Yoshida, H. Nishide,
`T. Yabu, Y. Kodama, T. Nakano, M. Taguchi, “3-
`Dimensional Stacked Capacitor Cell for 16M and 64M
`DRAMs”, IEDM Tech. Dig., pp. 592-595, Dec. 1988.
`[14] W. Wakamiya, Y. Tanaka, H. Kimura, H. Miyatake and
`S. Satoh, “Novel Stacked Capacitor Cell for 64Mb
`DRAM”, Symposium on VLSI Technology Dig. Tech.
`Papers, pp. 69-70, 1989.
`[15] H. Watanabe, T. Tatsumi, S. Ohnishi, T. Hamada, I.
`Honma, T. Kikkawa, “A New Cylindrical Capacitor
`Using Hemispherical Grained Si (HSG-Si) for 256Mb
`DRAMs,” IEDM Tech. Dig., pp. 259 - 262, Dec. 1992.
`[16] K.W. Lee, T. Nakamura, T. Ono, Y. Yamada, T.
`Mizukusa, H. Hashimoto, K.T. Park, H. Kurino and M.
`Koyanagi, “Three-Dimensional Shared Memory Fabri-
`cated Using Wafer Stacking Technology”, IEDM Tech.
`Dig., pp. 165-168, Dec. 2000.
`[17] T. Fukushima, Y. Yamada, H. Kikuchi, and M. Koy-
`anagi, “New Three-Dimensional Integration Technolo-
`gy Using Self-Assembly Technique”, IEDM Tech. Dig.,
`pp. 359-362, Dec. 2005.
`[18] M. Koyanagi, “Roadblocks in Achieving Three-Dimen-
`sional LSI”, 8th Symposium on Future Electron
`Devices Tech. Dig., pp.55-60, 1989.
`[19] H. Takata, T. Nakano, S. Yokoyama, S. Horiuchi, H.
`Itani, H. Tsukamoto and M. Koyanagi, “A novel fabri-
`cation technology for optically interconnected three-
`dimensional LSI by wafer aligning and bonding tech-
`nique,” Proc. Int. Semiconductor Device Research
`Symposium, pp.327-330, 1991.
`[20] M. Koyanagi, T. Nakamura, Y. Yamada, H. Kikuchi, T.
`Fukushima, T. Tanaka, and H. Kurino, “Three-Dimen-
`sional Integration Technology Based on Wafer Bond-
`ing With Vertical Buried Interconnections”, IEEE Trans.
`
`on Electron Devices, vol.53, pp. 2799-2808, 2006.
`[21] T. Ono, T. Mizukusa, T. Nakamura, Y. Yamada, Y.
`Igarashi, T. Morooka, H. Kurino and M. Koyanagi,
`“Three-Dimensional Processor System Fabricated by
`Wafer Stacking Technology”, IEEE COOL Chips V
`Tech. Dig., pp.186-193, 2002.
`[22] T. Fukushima, Y. Yamada, H. Kikuchi, T. Konno, J.
`Liang, K. Sasaki, K. Inamura, T. Tanaka, and M. Koy-
`anagi, “New Three-Dimensional Integration Technolo-
`gy Based on Reconfigured Wafer-on-Wafer Bonding
`Technique”, IEDM Tech. Dig., pp. 985-988, Dec. 2007.
`
`About the Author
`Mitsumasa Koyanagi was born in
`1947 in Hokkaido, Japan. He received
`the B.S. degree from Department of
`Electrical Engineering, Muroran Insti-
`tute of Technology, Japan in 1969 and
`the M.S. and Ph.D. degrees from
`Department of Electronic Engineering,
`Tohoku University in 1971 and 1974, respectively.
`
`He joined the Central Research Laboratory, Hitachi
`Co. Ltd. in 1974 where he had engaged in the
`research and development of DRAM and ASIC
`process and device technologies and invented a
`Stacked Capacitor DRAM memory cell which has
`been widely used in DRAM production. In 1985, he
`joined Xerox Palo Alto Research Center, where he
`was responsible for research on submicron CMOS
`devices, poly-Si TFT devices and analog/digital sensor
`LSI design. In 1988, he became a professor in the
`Research Center for Integrated Systems, Hiroshima
`University, Japan where he engaged in the research of
`sub-0.1um device fabrication and characterization,
`device modeling, poly-Si TFT devices, 3-D integration
`technology, optical interconnection and parallel com-
`puter system. He proposed a 3-D integration technol-
`ogy based on wafer-to-wafer bonding and through-Si
`vias (TSVs) in 1989. Since 1994, he has been a pro-
`fessor in the Intelligent System Design Lab., Depart-
`ment of Machine Intelligence and Systems Engineer-
`ing, and is currently a member of the Department of
`Bioengineering and Robotics, Graduate School of
`Engineering, Tohoku University, Japan. His current
`interests are 3-D integration technology, optical inter-
`connection, nano-CMOS devices, memory devices,
`parallel computer system specific for scientific com-
`putation, real-time image processing system, artificial
`retina chip and retinal prosthesis chip, brain-machine
`interface (BMI) and neural prosthesis chip, brain-like
`computer system.
`
`Dr. Koyanagi was awarded the IEEE Jun-Ichi Nishiza-
`wa Medal (2006), the IEEE Cledo Brunetti Award
`(1996) and Japan’s Ministry of Education, Culture,
`Sports, Science and Technology Award (2002), in
`addition to the Ohkouchi Prize (1992), SSDM (Solid-
`State Devices and Materials Conf.) Award (1994) and
`the Opto-Electronic Integration Technology (Izuo
`Hayashi) Award in 2004. He is an IEEE fellow.
`
`Winter 2008
`
`IEEE SSCS NEWS
`
`41
`
`Page 5 of 5