Pub Date : 2019-10-01DOI: 10.1109/3DIC48104.2019.9058795
J. Tada, Kazuto Takahashi, Ryusuke Egawa
This paper focuses on the convolutional neural network (CNN) accelerator with a three-dimension stacking integrated circuit (3DSIC) technology. In order to implement a CNN accelerator into several layers by using 3DSIC technology, circuit designers should consider several problems such as how many floating-point units and register-files implement on one layer, and how to place these. Because of these problems will affect the performance of the 3-D stacked CNN accelerator. This paper proposes a design scheme of the module for a 3-D stacked convolutional neural network accelerator, which consists of several half-precision floating-point fused multiply-add units and weight register-files. The proposed scheme achieves up to a 10% power consumption reduction compared to a 2-D implemented module when the number of units and register-files is sixteen, and the number of layers is two.
{"title":"A Design Scheme for 3-D Stacked CNN Accelerators","authors":"J. Tada, Kazuto Takahashi, Ryusuke Egawa","doi":"10.1109/3DIC48104.2019.9058795","DOIUrl":"https://doi.org/10.1109/3DIC48104.2019.9058795","url":null,"abstract":"This paper focuses on the convolutional neural network (CNN) accelerator with a three-dimension stacking integrated circuit (3DSIC) technology. In order to implement a CNN accelerator into several layers by using 3DSIC technology, circuit designers should consider several problems such as how many floating-point units and register-files implement on one layer, and how to place these. Because of these problems will affect the performance of the 3-D stacked CNN accelerator. This paper proposes a design scheme of the module for a 3-D stacked convolutional neural network accelerator, which consists of several half-precision floating-point fused multiply-add units and weight register-files. The proposed scheme achieves up to a 10% power consumption reduction compared to a 2-D implemented module when the number of units and register-files is sixteen, and the number of layers is two.","PeriodicalId":440556,"journal":{"name":"2019 International 3D Systems Integration Conference (3DIC)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126619115","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-10-01DOI: 10.1109/3DIC48104.2019.9058779
A. Noriki, I. Tamai, Y. Ibusuki, A. Ukita, S. Suda, D. Shimura, Y. Onawa, H. Yaegashi, T. Amano
Optical through Si vias (TSVs) for Si photonics have been studied for many new applications and grating couplers have been used as vertical optical I/O (input/output) devices. Recently, we demonstrated a new vertical optical I/O using an integrated curved micro-mirror. Compared to grating couplers, broadband optical I/O can be available. In this work, a feasibility study of an optical TSV using the curved micromirror was carried out. A curved micro-mirror designed for an optical TSV of Si interposer up to 200-micrometer thickness was integrated on a Si photonics chip and the vertical optical output from the Si waveguide was demonstrated.
{"title":"Optical TSV Using Si-Photonics Integrated Curved Micro-Mirror","authors":"A. Noriki, I. Tamai, Y. Ibusuki, A. Ukita, S. Suda, D. Shimura, Y. Onawa, H. Yaegashi, T. Amano","doi":"10.1109/3DIC48104.2019.9058779","DOIUrl":"https://doi.org/10.1109/3DIC48104.2019.9058779","url":null,"abstract":"Optical through Si vias (TSVs) for Si photonics have been studied for many new applications and grating couplers have been used as vertical optical I/O (input/output) devices. Recently, we demonstrated a new vertical optical I/O using an integrated curved micro-mirror. Compared to grating couplers, broadband optical I/O can be available. In this work, a feasibility study of an optical TSV using the curved micromirror was carried out. A curved micro-mirror designed for an optical TSV of Si interposer up to 200-micrometer thickness was integrated on a Si photonics chip and the vertical optical output from the Si waveguide was demonstrated.","PeriodicalId":440556,"journal":{"name":"2019 International 3D Systems Integration Conference (3DIC)","volume":"362 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115985056","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-10-01DOI: 10.1109/3DIC48104.2019.9058844
M. Nomura
Heat transfer in Si nanostructure is not only an interesting topic in fundamental physics, but also an important practical study for efficient heat dissipation in electronic devices. In this paper, we discuss characteristic thermal phonon transport property in nanostructured Si membranes, where the dimension is smaller than phonon mean free path. We mainly focus on heat flux control technique in Si membranes using ballisticity and wave nature of phonons, which may be useful for deeper understanding of thermal dissipation in electronic devices with nanostructures.
{"title":"Heat Transfer in Nanostructured Si and Heat Flux Control Technique","authors":"M. Nomura","doi":"10.1109/3DIC48104.2019.9058844","DOIUrl":"https://doi.org/10.1109/3DIC48104.2019.9058844","url":null,"abstract":"Heat transfer in Si nanostructure is not only an interesting topic in fundamental physics, but also an important practical study for efficient heat dissipation in electronic devices. In this paper, we discuss characteristic thermal phonon transport property in nanostructured Si membranes, where the dimension is smaller than phonon mean free path. We mainly focus on heat flux control technique in Si membranes using ballisticity and wave nature of phonons, which may be useful for deeper understanding of thermal dissipation in electronic devices with nanostructures.","PeriodicalId":440556,"journal":{"name":"2019 International 3D Systems Integration Conference (3DIC)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129902348","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-10-01DOI: 10.1109/3DIC48104.2019.9058840
Sreejith Kochupurackal Rajan, M. Li, M. Bakir, G. May
In this work, we propose and demonstrate the use of metal electroless plating along with mechanical self-alignment as a method to create high density scalable interconnects between chiplets using a low temperature process flow. Void free and selective bonding of a copper pillar array at 50 ^m pitch using electroless nickel deposition is demonstrated at a temperature of 95°C and atmospheric pressure. Mechanical self-alignment using precision ruby balls is demonstrated to attain alignment accuracies within 2 μm.
{"title":"High Density and Low-Temperature Interconnection Enabled by Mechanical Self-Alignment and Electroless Plating","authors":"Sreejith Kochupurackal Rajan, M. Li, M. Bakir, G. May","doi":"10.1109/3DIC48104.2019.9058840","DOIUrl":"https://doi.org/10.1109/3DIC48104.2019.9058840","url":null,"abstract":"In this work, we propose and demonstrate the use of metal electroless plating along with mechanical self-alignment as a method to create high density scalable interconnects between chiplets using a low temperature process flow. Void free and selective bonding of a copper pillar array at 50 ^m pitch using electroless nickel deposition is demonstrated at a temperature of 95°C and atmospheric pressure. Mechanical self-alignment using precision ruby balls is demonstrated to attain alignment accuracies within 2 μm.","PeriodicalId":440556,"journal":{"name":"2019 International 3D Systems Integration Conference (3DIC)","volume":"154 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121696507","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-10-01DOI: 10.1109/3dic48104.2019.9058843
Rui Liang, Sungho Lee, Y. Miwa, Kousei Kumahara, M. Murugesan, H. Kino, T. Fukushima, Tetsu Tanaka
Through-silicon vias (TSVs) is one of the key technologies for 3D integration. To solve the issues induced by the high-temperature process for TSV liner formation in the multichip-to-wafer (MCtW) process, we applied the low-temperature SiO2 deposition method called OER (Ozone-Ethylene Radical generation)-TEOS-CVD®. In this study, we fabricated the MIS capacitors with the TSV liner deposited by OER-TEOS-CVD® at 150°C and room temperature (RT), and compared both the coverage and electrical characteristics with that formed by conventional plasma-enhanced chemical vapor deposition (PE-CVD) at 200°C. Furthermore, we analyzed these SiO2liners by FTIR and synchrotron XPS. These results showed that the OER-TEOS-CVD® has high potentials to realize highly-reliable TSVs and to apply to various processes in 3D integration.
{"title":"Impacts of Deposition Temperature and Annealing Condition on Ozone-Ethylene Radical Generation-TEOS-CVD SiO2 for Low-Temperature TSV Liner Formation","authors":"Rui Liang, Sungho Lee, Y. Miwa, Kousei Kumahara, M. Murugesan, H. Kino, T. Fukushima, Tetsu Tanaka","doi":"10.1109/3dic48104.2019.9058843","DOIUrl":"https://doi.org/10.1109/3dic48104.2019.9058843","url":null,"abstract":"Through-silicon vias (TSVs) is one of the key technologies for 3D integration. To solve the issues induced by the high-temperature process for TSV liner formation in the multichip-to-wafer (MCtW) process, we applied the low-temperature SiO2 deposition method called OER (Ozone-Ethylene Radical generation)-TEOS-CVD®. In this study, we fabricated the MIS capacitors with the TSV liner deposited by OER-TEOS-CVD® at 150°C and room temperature (RT), and compared both the coverage and electrical characteristics with that formed by conventional plasma-enhanced chemical vapor deposition (PE-CVD) at 200°C. Furthermore, we analyzed these SiO2liners by FTIR and synchrotron XPS. These results showed that the OER-TEOS-CVD® has high potentials to realize highly-reliable TSVs and to apply to various processes in 3D integration.","PeriodicalId":440556,"journal":{"name":"2019 International 3D Systems Integration Conference (3DIC)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134220863","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-10-01DOI: 10.1109/3DIC48104.2019.9058903
Alit Apriyana Anak Agung, P. Zhao, C. S. Tan
This paper presents the design of TiN guard ring structure that is built around the through silicon via (TSV) to improve the signal integrity (SI) performance of parallel networking lines by suppressing the crosstalk from adjacent TSV. The guard ring enhances the insertion loss by up to 1.6 dB and the crosstalk isolation by 20-26 dB across a range of 0-40 GHz. The attainable data rate is higher than 50 Gbps under 50 Ohm load and higher than 10 Gbps under 1 pF capacitive loading. The obtained signal-to-noise ratio (SNR) is greater than 15.
{"title":"TiN Guard Ring Around TSV for Cross-Talk Suppression of Parallel Networking of Data Center","authors":"Alit Apriyana Anak Agung, P. Zhao, C. S. Tan","doi":"10.1109/3DIC48104.2019.9058903","DOIUrl":"https://doi.org/10.1109/3DIC48104.2019.9058903","url":null,"abstract":"This paper presents the design of TiN guard ring structure that is built around the through silicon via (TSV) to improve the signal integrity (SI) performance of parallel networking lines by suppressing the crosstalk from adjacent TSV. The guard ring enhances the insertion loss by up to 1.6 dB and the crosstalk isolation by 20-26 dB across a range of 0-40 GHz. The attainable data rate is higher than 50 Gbps under 50 Ohm load and higher than 10 Gbps under 1 pF capacitive loading. The obtained signal-to-noise ratio (SNR) is greater than 15.","PeriodicalId":440556,"journal":{"name":"2019 International 3D Systems Integration Conference (3DIC)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131352346","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-10-01DOI: 10.1109/3DIC48104.2019.9058900
K. Sakui, T. Ohba
This paper proposes a fundamental architecture for the High Bandwidth Memory (HBM) with the bumpless TSV for the Wafer-on-Wafer (WOW) technology. The bumpless interconnects technology can increase the number of TSVs per chip with fine pitch of TSVs, and reduce the impedance of the TSV interconnects with no bumps. Therefore, a further higher speed and higher density HBM can be realized. Also, the High Bandwidth NAND (HBN), which can read and program by plane instead of by line by using the bumpless TSV, has been proposed.
{"title":"High Bandwidth Memory (HBM) and High Bandwidth NAND (HBN) with the Bumpless TSV Technology","authors":"K. Sakui, T. Ohba","doi":"10.1109/3DIC48104.2019.9058900","DOIUrl":"https://doi.org/10.1109/3DIC48104.2019.9058900","url":null,"abstract":"This paper proposes a fundamental architecture for the High Bandwidth Memory (HBM) with the bumpless TSV for the Wafer-on-Wafer (WOW) technology. The bumpless interconnects technology can increase the number of TSVs per chip with fine pitch of TSVs, and reduce the impedance of the TSV interconnects with no bumps. Therefore, a further higher speed and higher density HBM can be realized. Also, the High Bandwidth NAND (HBN), which can read and program by plane instead of by line by using the bumpless TSV, has been proposed.","PeriodicalId":440556,"journal":{"name":"2019 International 3D Systems Integration Conference (3DIC)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133838362","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-10-01DOI: 10.1109/3DIC48104.2019.9058784
T. R. Harris, W. R. Davis, S. Lipa, W. S. Pitts, P. Franzon
This paper presents thermal measurement data of GaN HEMT on CMOS heterogeneous integration (HI) using a Diverse Accessible Heterogeneous Integration (DAHI) process. Thermal T3ster measurements, a product and service available from Mentor are presented. The method uses thermal transients to characterize the vertical thermal path stack including the package. Here the thermal dominance of the thermal interface at the die attachment is apparent. The T3ster measurements are contrasted with in-channel micro-Raman thermal measurements along with simulated results.
{"title":"Vertical Stack Thermal Characterization of Heterogeneous Integration and Packages","authors":"T. R. Harris, W. R. Davis, S. Lipa, W. S. Pitts, P. Franzon","doi":"10.1109/3DIC48104.2019.9058784","DOIUrl":"https://doi.org/10.1109/3DIC48104.2019.9058784","url":null,"abstract":"This paper presents thermal measurement data of GaN HEMT on CMOS heterogeneous integration (HI) using a Diverse Accessible Heterogeneous Integration (DAHI) process. Thermal T3ster measurements, a product and service available from Mentor are presented. The method uses thermal transients to characterize the vertical thermal path stack including the package. Here the thermal dominance of the thermal interface at the die attachment is apparent. The T3ster measurements are contrasted with in-channel micro-Raman thermal measurements along with simulated results.","PeriodicalId":440556,"journal":{"name":"2019 International 3D Systems Integration Conference (3DIC)","volume":"34 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117274566","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-10-01DOI: 10.1109/3DIC48104.2019.9058832
T. Miura, M. Sakakibara, H. Takahashi, T. Taura, K. Tatani, Y. Oike, T. Ezaki
In this paper, we propose a 3D stacked global shutter CMOS image sensor with 3M Cu-Cu connections. Using a fine pitch and a large number of Cu-Cu connection technology, we achieved 1.46M pixels of size 6.9 μm × 6.9 μm. The pixel evaluation results reveal that all the 3M Cu-Cu connections were realized without defect.
{"title":"A 6.9 μm Pixel-Pitch 3D Stacked Global Shutter CMOS Image Sensor with 3M Cu-Cu connections","authors":"T. Miura, M. Sakakibara, H. Takahashi, T. Taura, K. Tatani, Y. Oike, T. Ezaki","doi":"10.1109/3DIC48104.2019.9058832","DOIUrl":"https://doi.org/10.1109/3DIC48104.2019.9058832","url":null,"abstract":"In this paper, we propose a 3D stacked global shutter CMOS image sensor with 3M Cu-Cu connections. Using a fine pitch and a large number of Cu-Cu connection technology, we achieved 1.46M pixels of size 6.9 μm × 6.9 μm. The pixel evaluation results reveal that all the 3M Cu-Cu connections were realized without defect.","PeriodicalId":440556,"journal":{"name":"2019 International 3D Systems Integration Conference (3DIC)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131026198","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-10-01DOI: 10.1109/3DIC48104.2019.9058846
Q. Dinh, K. Kondo, T. Hirato
The mismatch in thermal expansion coefficient (TEC) between copper and silicon causes serious problems in three-dimensional (3D) packaging. The common problem is TSV pumping when TSV is exposed to high temperature (400oC-600oC) during the wiring process. The copper pumping destroys wiring above TSV and leads to the failure of electronic devices. Other problem is the area on silicon around the TSV where the transistors cannot be formed due to stress caused by copper when annealing. With our low TEC additive (additive A), copper pumping height in 5× 20 μm p-TEOS TSV was reduced to 0.5 μm from 2.0 μm in case of conventional copper. We also investigated the effect of polyimide which is used as liner layer in the TSV on copper pumping reduction. The first screening result showed that the pumping height of conventional copper in polyimide TSV was only 0.8 μm., compared to 2.0 μm p-TEOS TSV and 1.2 μm O3-TEOS TSV.
{"title":"Reduction of TSV Pumping","authors":"Q. Dinh, K. Kondo, T. Hirato","doi":"10.1109/3DIC48104.2019.9058846","DOIUrl":"https://doi.org/10.1109/3DIC48104.2019.9058846","url":null,"abstract":"The mismatch in thermal expansion coefficient (TEC) between copper and silicon causes serious problems in three-dimensional (3D) packaging. The common problem is TSV pumping when TSV is exposed to high temperature (400oC-600oC) during the wiring process. The copper pumping destroys wiring above TSV and leads to the failure of electronic devices. Other problem is the area on silicon around the TSV where the transistors cannot be formed due to stress caused by copper when annealing. With our low TEC additive (additive A), copper pumping height in 5× 20 μm p-TEOS TSV was reduced to 0.5 μm from 2.0 μm in case of conventional copper. We also investigated the effect of polyimide which is used as liner layer in the TSV on copper pumping reduction. The first screening result showed that the pumping height of conventional copper in polyimide TSV was only 0.8 μm., compared to 2.0 μm p-TEOS TSV and 1.2 μm O3-TEOS TSV.","PeriodicalId":440556,"journal":{"name":"2019 International 3D Systems Integration Conference (3DIC)","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114802526","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}