• JSTS:Journal of Semiconductor Technology and Science
• /
• v.3 no.2
• /
• pp.76-82
• /
• 2003

The robust stack storage node and sufficient cell capacitance for high performance is indispensable for 90 nm DRAM capacitor. For the first time, we successfully demonstrated MIS capacitor process integration for 90 nm DRAM technology. Novel cell layout and integration technology of 90 nm DRAM capacitor is proposed and developed, and it can be extended to the next generation DRAM. Diamond-shaped OCS with 1.8 um stack height is newly developed for large capacitor area with better stability. Furthermore, the novel $Al_2O_3/HfO_2$ dielectric material with equivalent oxide thickness (EOT) of 25 ${\AA}$ is adopted for obtaining sufficient cell capacitance. The reliable cell capacitance and leakage current of MIS capacitor is obtained with ~26 fF/cell and < 1 fA/ceil by $Al_2O_3/HfO_2$ dielectric material, respectively.

• Journal of IKEEE
• /
• v.25 no.1
• /
• pp.180-192
• /
• 2021

One widely used algorithm for image recognition and pattern detection is the convolution neural network (CNN). To efficiently handle convolution operations, which account for the majority of computations in the CNN, we use hardware accelerators to improve the performance of CNN applications. In using these hardware accelerators, the CNN fetches data from the off-chip DRAM, as the massive computational volume of data makes it difficult to derive performance improvements only from memory inside the hardware accelerator. In other words, data communication between off-chip DRAM and memory inside the accelerator has a significant impact on the performance of CNN applications. In this paper, a simulator for the CNN is developed to analyze the main memory or DRAM access with respect to the size of the on-chip memory or global buffer inside the CNN accelerator. For AlexNet, one of the CNN architectures, when simulated with increasing the size of the global buffer, we found that the global buffer of size larger than 100kB has 0.8x as low a DRAM access count as the global buffer of size smaller than 100kB.

• Proceedings of the IEEK Conference
• /
• 2007.07a
• /
• pp.433-434
• /
• 2007

This paper describes a low-voltage dynamic random-access memory (DRAM) focusing on subthreshold leakage reduction during self-refresh (sleep) mode. By sharing a power switch, multiple iterative circuits such as row and column decoders have a significantly reduced subthreshold leakage current. To reduce the leakage current of complex logic gates, dual channel length scheme and input vector control method are used. Because all node voltages during the standby mode are deterministic, zigzag super-cutoff CMOS is used, allowing to Preserve internal data. MTCMOS technique Is also used in the circuits having no need to preserve internal data. Sub-1.2-V 1-Gb mobile DDR DRAM employing all these low-power techniques was designed in a 60 nm CMOS technology and achieved over 77% reduction of overall leakage current during the self-refresh mode.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.7 no.4
• /
• pp.833-837
• /
• 2012

This paper proposes a DRAM data retention time enhancement method that minimizes silicon loss and undercut at STI sidewall by reducing the SC1 (Standard Cleaning) time. SC1 time optimization debilitates the parasitic electric field in STI's top corner, which reduces an inverse narrow width effect to result in reduction of channel doping density without increasing the subthreshold leakage of cell Tr. Moreover, it minimizes the electric field in depletion area from cell junction to P-well, increasing yield or data retention time.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.37 no.7
• /
• pp.7-16
• /
• 2000

This paper reports a methodology and its application for extracting cell capacitances and parasitic capacitances in a stacked DRAM cell structure by a numerical technique. To calculate the cell and parasitic capacitances, we employed finite element method (FEM), The three-dimensional DRAM cell structure is generated by solid modeling based on two-dimensional mask layout and transfer data. To obtain transfer data for generating three-dimensional simulation structure, topography simulation is performed. In this calculation, an exemplary structure comprising 4 cell capacitors with a dimension of $2.25{\times}1.75{\times}3.45{\mu}m^3$, 70,078 nodes with 395,064 tetrahedra were used in ULTRA SPARC 10 workstation. The total CPU time for the simulation was about 25 minutes, while the memory size of 201MB was required. The calculated cell capacitance is 24.34fF per cell, and the influential parasitic capacitances in a stacked DRAM cell are investigated.

• Proceedings of the Korean Institute of Communication Sciences Conference
• /
• 1986.04a
• /
• pp.234-238
• /
• 1986

• Journal of the Korean Ceramic Society
• /
• v.34 no.10
• /
• pp.1060-1066
• /
• 1997

In this paper, the effects of La addition of PLZT thin film prepared by sol-gel method on the capacitor characteristics are investigated for gigabit generation DRAM applications. The addition of La on the PLZT capacitor results in a trade-off between charge storage density(Qc') and leakage current density(Jl). As La content increases, Qc' and permeability(εr) at 0V are reduced while Jl is significantly decreased. It is demonstrated that 5% La doping of PZT can substantially reduce Jl and also improve resistance to fatigue while incurring only minimal degradation of Qc'. Very low leakage current density (5×10-7 A/㎠ even at 125℃) and high charge storage density (100fC/㎛2) under VDD/2=1V conditions are achieved using 5% La doped PZT thin films for gigabit DRAM capacitor dielectrics. In addition, the fatigue and TDDB measurements indicate good reliability of the PLZT capacitors.

• Proceedings of the IEEK Conference
• /
• 2006.06a
• /
• pp.485-486
• /
• 2006

This paper proposes a novel low-cost CMOS temperature sensor for controlling the self-refresh period of a mobile DRAM. In this temperature sensor, ring oscillators composed of cascaded inverter stages are used to obtain the temperature of the chip. This method is highly area-efficient, simple and easy for IC implementation as compared to traditional temperature sensors based on analog bandgap reference circuits. The proposed CMOS temperature sensor was fabricated with 80 nm 3-metal DRAM process. It occupies a silicon area of only about less than $0.02\;mm^2$ at $10^{\circ}C$ resolution with under 5uW power consumption at 1 sample/s processing rate. This area is about 33% of conventional temperature sensor in mobile DRAM.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.39 no.11
• /
• pp.1-6
• /
• 2002

This study is about optimization of Laser Repair Condition in EDS Line to improve AC and DC characteristic of high speed DRAM. The margin of AC parameter can be improved by forcing the proper DC generator levels and also improved by cutting the optional fuse about characteristics.

• Electrical & Electronic Materials
• /
• v.8 no.5
• /
• pp.648-657
• /
• 1995

정보처리의 다양화, 고속화를 위하여 장래의 집적회로는 다량의 정보를 단시간에 처리하지 않으면 안된다. 종래, 3년에 4배의 고집적화가 실현되어 LSI개발에 기술 견인차의 역할을 하고 있는 DRAM(Dynamic Random Access Memory)은 미세화기술의 한계를 우려하면서도 오히려 개발에 박차를 가하고 있다. 이러한 DRAM의 미세, 대용량화에는 미세가공 기술, 새로운 메모리 셀과 트랜지스터 기술, 새로운 회로 기술, 그 이외에 재료박막기술, Computer aided design/Design automation(CAD/DA) 기술, 검사평가기술 혹은 소형팩키지(package)기술등의 광범위한 기술발전이 뒷받침되어 왔다. 그 중에서 미세가공 기술 및 새로운 트랜지스터 기술과 메모리 셀 기술을 중심으로 개발 동향을 살펴보고 최근에 발표된 1Gbit DRAM의 시제품 기술에 대하여 분석해 보기로 한다.