Back-end Processing Research Articles

ImgLib2 for large scale image analysis and visualization

Event Abstract Back to Event ImgLib2 for large scale image analysis and visualization Stephan Saalfeld1*, Tobias Pietzsch1, Stephan Preibisch2, 3 and Pavel Tomancak1 1 Max Planck Institute of Molecular Cell Biology and Genetics, Germany 2 Albert Einstein College of Medicine, United States 3 Max Planck Institute of Molecular Cell Biology and Genetics, Germany Today, both connectivity and detailed neuroanatomy of biological nervous tissue are reconstructed from light- and electron-microscopy images. Structures in neuronal tissue span a wide range of scales, they have both fine details and large extent. Accordingly, large volumes need to be imaged at very high resolution resulting in image data of overwhelming and ever increasing size. New approaches to both algorithm development for image analysis and data storage and access management are desperately required. We propose our Java library ImgLib2 for n-dimensional data representation and manipulation [1] as a valuable tool in this context. ImgLib2 separates pixel-algebra, data access and data representation in memory by virtualization. It makes algorithm development independent of infrastructure design and vice versa, simplifying both lines of development. In consequence, it is easier to express new or existing approaches for image analysis in software. New data sources are immedately available to existing algorithm implementations. ImgLib2 collaborates seamlessly with existing infrastructure. It is typically sufficient to implement one basic accessor to connect an existing data source to the library. We demonstrate ImgLib2's flexibility in two distinct and relevant contexts: (1) a rich-client visualization and annotation tool for remotely stored image volumes of many terabytes size [2], and (2) as a server-side image processing backend for the browser-based collaborative annotation tool CATMAID [3].

The All-Optical Fiber Perimeter Alarm System

A all-fiber perimeter alarm system is designed which is composed of the optical Fiber Sensor Module(FSM) laid along the perimeter and Security Information Control Center(SICC) (including display terminals-the large screen display, the terminal of personnel operation, the system control host, etc.) and optical fiber fence (or sensor board). It uses optical fiber (cable) as the sensing and transmission device, through directly contacting with optical fiber (cable) or through the carrying material, passing various disturbances to the fiber (cable), carrying continuous real-time monitoring and collecting disturbance data. Through the back-end analysis, processing and intelligent identification, it can determine the different types of external disturbances. The system can automatically obtain the real-time alarm information that the optical cable generates and process these warning messages according to predefined business logic, then generate sound and light warning prompt and provide several functions of statistics, query on the warning message.

Resistive Switching in Organic Memory Device Based on Parylene-C With Highly Compatible Process for High-Density and Low-Cost Memory Applications

In this paper, the resistive switching behavior of a single-component-polymer resistive memory device based on polychloro-para-xylylene (parylene-C) is comprehensively studied. With the excellent chemical stability and high process compatibility of parylene-C, an 8 × 8 crossbar array with a sandwiched structure of parylene-C and active electrodes is fabricated, which can be integrated in the CMOS back-end process and shows great potentials for future transparent, low-cost, flexible, and high-density nonvolatile memory applications. This organic memory device exhibits excellent performance with a 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">7</sup> ON/OFF current ratio, nanosecond set/reset speed, and low switching voltages, as well as good retention and cycling endurance behaviors. The switching mechanism is systematically investigated with the comparison between active electrodes (Al, Cu, or Ag) and an inert electrode (Pt), as well as the dependence on temperature and device area.

Depicting the evolving scenario of translational-guided drug development

The development process of most important targeted agents currently used in oncologic clinical practice has unveiled the fact that using these types of drugs efficaciously is best supported by the availability of appropriate predictive biomarkers. Conversely, the magnitude of the effect obtained from administering such drugs to patients that respond to them on the basis of known biomarkers is likely to be diluted by concomitantly administering the drugs to an unselected patient population as well. This scenario could compromise the future of bona fide compounds due to the bias of non-selection when measuring drug benefit. Usually this agent-biomarker matching process and its approval based on positive indications of spent takes in identifying a rational biomarker, and then the back end process of validating the assay used to measure it. In this regard, Bando and collaborators compare, in this issue of the journal, the three methods for determining KRAS mutation in colorectal carcinoma with the goal of delineating the best test for this purpose. The history of the role of KRAS mutations in metastatic colorectal carcinoma well illustrates the diverse processes underlying predictive biomarker identification for antineoplastic targeted agents during the last 20 years, as reviewed in Table 1. Mutations in codons 12, 13, 61 and 126 of the KRAS protein were accepted as determinants of resistance to treatment with anti-epidermal growth factor receptor monoclonal antibodies (anti-EGFR MoAbs) in colorectal carcinoma, years after cetuximab was first granted approval for its treatment by regulatory agencies [1, 2]. At that time, the development of anti-EGFR MoAbs was focused on EGFR amplification and overexpression as the primary characteristics for selecting patients for clinical trials [3, 4]. The discovery of the role of KRAS mutations redirected the development plans for these compounds, causing some ongoing clinical trials to be amended to exclude patients with KRAS activating mutations. When this new approach was instituted, clinical endpoints were successfully met, leading these drugs to broad approval in many clinical settings of colorectal carcinoma. This example and other successes resulting from the correct implementation of translational research data in clinical trials have shifted the panorama of drug development (Table 1). Accordingly, we have witnessed in the last decade clinical and translational research in oncology being brought closer together in a process intended to develop in parallel the best targeted agents and to align them with the best predictive biomarkers to guide their use starting with in early phases of the drug development process. This rapprochement bridging the gap between research and clinical application has modified both arenas. Such alignment is shown not only in the increasing number of research institutes linked to health care facilities but also by the changes that phase 1 clinical trial units have undergone [5]. In keeping with this trend, the search for biomarkers of drug effect (pharmacodynamic markers) and identification of potential target subpopulations of patients that, until recently, has occurred only in the last phases of drug G. Argiles (&) J. Rodon J. Tabernero Medical Oncology Department, Vall d’Hebron University Hospital, Barcelona, Spain e-mail: gargiles@vhebron.net

FPGA-based gating and logic for multichannel single photon counting

We present results characterizing multichannel InGaAs single photon detectors utilizing gated passive quenching circuits (GPQC), self-differencing techniques, and field programmable gate array (FPGA)-based logic for both diode gating and coincidence counting. Utilizing FPGAs for the diode gating frontend and the logic counting backend has the advantage of low cost compared to custom built logic circuits and current off-the-shelf detector technology. Further, FPGA logic counters have been shown to work well in quantum key distribution (QKD) test beds. Our setup combines multiple independent detector channels in a reconfigurable manner via an FPGA backend and post processing in order to perform coincidence measurements between any two or more detector channels simultaneously. Using this method, states from a multi-photon polarization entangled source are detected and characterized via coincidence counting on the FPGA. Photons detection events are also processed by the quantum information toolkit for application testing (QITKAT).

Hot embossing of thick amorphous fluoropolymer for back end processing of infrared arrays

A process is presented for patterning vias into thick amorphous fluoropolymer layers for a novel In bump fabrication process, achieved using a hot embossing technique. The technique uses a patterned Si stamp that employs a two-step etching process to obtain pillars with a controlled positive sidewall angle. After embossing with the Si stamp, vias are formed in amorphous fluoropolymer layers. A postembossing blanket reactive ion etch step is then used to remove excess fluoropolymer from the bottoms of the vias, exposing a Ni film. Successful electroplating of In bumps into vias initiated at the Ni layer is demonstrated, confirming complete removal of excess fluoropolymer.

Multilayer Perceptron Classification of Unknown Volatile Chemicals from the Firing Rates of Insect Olfactory Sensory Neurons and Its Application to Biosensor Design

In this letter, we use the firing rates from an array of olfactory sensory neurons (OSNs) of the fruit fly, Drosophila melanogaster, to train an artificial neural network (ANN) to distinguish different chemical classes of volatile odorants. Bootstrapping is implemented for the optimized networks, providing an accurate estimate of a network's predicted values. Initially a simple linear predictor was used to assess the complexity of the data and was found to provide low prediction performance. A nonlinear ANN in the form of a single multilayer perceptron (MLP) was also used, providing a significant increase in prediction performance. The effect of the number of hidden layers and hidden neurons of the MLP was investigated and found to be effective in enhancing network performance with both a single and a double hidden layer investigated separately. A hybrid array of MLPs was investigated and compared against the single MLP architecture. The hybrid MLPs were found to classify all vectors of the validation set, presenting the highest degree of prediction accuracy. Adjustment of the number of hidden neurons was investigated, providing further performance gain. In addition, noise injection was investigated, proving successful for certain network designs. It was found that the best-performing MLP was that of the double-hidden-layer hybrid MLP network without the use of noise injection. Furthermore, the level of performance was examined when different numbers of OSNs used were varied from the maximum of 24 to only 5 OSNs. Finally, the ideal OSNs were identified that optimized network performance. The results obtained from this study provide strong evidence of the usefulness of ANNs in the field of olfaction for the future realization of a signal processing back end for an artificial olfactory biosensor.

Gas sensing properties of novel CuO nanowire devices

We report on novel gas sensing devices based on cupric oxide (CuO) nanowires which are synthesized on-chip by thermal oxidation of electroplated copper microstructures. This technique enables the direct integration of a multitude of CuO nanowires, which bridge the electrical contacts of a conductometric gas sensor. The CuO nanowire bridges exhibit a huge surface-to-volume ratio and are entirely surrounded by the gas atmosphere, which is a highly favorable gas sensor configuration. As a result, the CuO nanowire gas sensor devices are able to detect carbon monoxide (CO) down to a concentration of 10ppm and exhibit extraordinary sensitivity to hydrogen sulfide (H2S) where concentrations down to 10ppb have been detected, even in the presence of humidity. For characterization of the CuO nanowires, X-ray diffraction measurements, transmission electron microscopy and electron energy loss spectroscopy are employed. As no process temperatures higher than 400°C are required for the fabrication of the CuO nanowire devices, our approach can be employed in a CMOS backend process enabling the realization of a fully silicon integrated CuO nanowire gas sensing device.

SiCer - A substrate to combine ceramic and silicon based micro systems

A silicon-on-ceramic (SiCer) wafer-shaped substrate, allowing a simultaneous fabrication of ceramic and silicon based microsystems is presented. The substrate, based on the recently introduced 'silicon-on-ceramic integration concept', can be processed by standard MEMS and LTCC technologies. The monolithic compound is fabricated by using lamination and pressure assisted firing of a nano patterned silicon wafer with a low temperature cofired ceramic tape (LTCC). The LTCC is adapted to silicon (TCE and morphology). LTCC functionalities such as electrical and thermal vias, passives, fluidic channels, etc.,. can be pre-processed in grid dimensions or individually. Through Silicon Vias (TSV) in the silicon layer can be combined with the vias in the LTCC. The silicon layer can be fabricated as thin as necessary for the desired functionality, whereas the insulating LTCC layer ensures the mechanical stability of the SiCer wafer. In addition, the LTCC layer includes electrical functionalities like signal routing and passive integration. Using SiCer substrates, the bigger part of back-end processes for silicon microsystems such as assembly and the crucial part of packaging, can take place before any micro structuring is generated. Thus, the influence of these processes to the sensitive micro structures can be minimized. The SiCer substrate is fundamentally temperature-stable up to 1000 °C. The fabrication steps of the SiCer substrate, particularly several options for TSV's are explained. Material characteristics and new technological approaches with the SiCer substrate are presented.

Catastrophic flip-chip failures at thermal cycles caused by micro-cracks in passivation layer, present only in the spacing between minimum width stripes of last metal level

Our case study concerns flip-chip devices, assembled on Printed Circuit Board with direct bumping, failing catastrophically under thermal cycles stress. The failure analysis highlighted a deep crack in silicon, systematically located under a particular bump. We show the path followed to understand failure mechanism by means analyses performed at various back-end process steps, up to wafer inspection. Micro-cracks in passivation layer were found, generated only in the spacing between minimum width stripes of last metal layer. Fixed the multiple passivation layer, necessary to protect the devices up to bumping process manufacturer, layout study and 3D simulator stress evaluation, allowed us to identify the metal stripe width as the “key” factor.The implemented solution was minimum metal 4 stripes widening. The devices with new layout were submitted to thermal cycles stress, resulting in zero failures (up to 500 cycles), versus 30% of failure rate (after 100 cycles) of first layout version.

Opening Up Science: Towards an Agenda of Open Science in Industry and Academia

The shift towards open innovation has substantially changed the scientific and practical perception of corporate innovation. While scientific studies on open innovation are burgeoning, present research underlies a business-centric view that has focused on the back-end of the innovation process. The impact and implications of open innovation on academic and industrial science at the very front-end of the innovation process have so far been neglected. Our paper presents a conceptualization of open science and research as a peculiarity under the roof of open innovation. We propose four perspectives, outline current trends, and present directions for future developments.

Effect of Prebonding Anneal on the Microstructure Evolution and Cu–Cu Diffusion Bonding Quality for Three-Dimensional Integration

Electroplated copper (Cu) films are often annealed during back-end processes to stabilize grain growth in order to improve their electrical properties. The effect of prebonding anneal and hence the effective initial grain size of the Cu films on the final bond quality are studied using a 300-nm-thick Cu film that was deposited on a 200-mm silicon (Si) wafer and bonded at 300°C. As compared with the control wafer pair with a prebonding anneal at 300°C for 1 h in N2, the wafer pair without a prebonding anneal showed greater improvement in void density based on c-mode scanning acoustic microscopy (c-SAM). Dicing yield and shear strength were also enhanced when a prebonding anneal was not applied. This improvement is due to substantial grain growth of smaller Cu grains during the bonding process, which leads to a stronger Cu–Cu bond. Our work has identified a Cu–Cu bonding process with a lower total thermal budget, which is seen as a favorable option for future three-dimensional (3D) integrated circuit (IC) technology.

A physiologically inspired model of auditory stream segregation based on a temporal coherence analysis

The ability to perceptually separate acoustic sources and focus one's attention on a single source at a time is essential for our ability to use acoustic information. In this study, a physiologically inspired model of human auditory processing Jepsenetal., 2008 was used as a front end of a model for auditory stream segregation. A temporal coherence analysis Elhilalietal., 2009 was applied at the output of the preprocessing, using the coherence across tonotopic channels to group activity across frequency. Using this approach, the described model is able to quantitatively account for classical streaming phenomena relying on frequency separation and tone presentation rate, such as the temporal coherence boundary and the fission boundary vanNoorden, 1975. The same model also accounts for the perceptual grouping of distant spectral components in the case of synchronous presentation. The most essential components of the front-end and back-end processing in the framework of the presented model are analyzed and future perspectives discussed.

Deuterium-incorporated gate oxide of MOS devices fabricated by using deuterium ion implantation

In the aspect of metal-oxide-semiconductor (MOS) device reliability, deuterium-incorporated gate oxide could be utilized to suppress the wear-out that is combined with oxide trap generation. An alternative deuterium process for the passivation of oxide traps or defects in the gate oxide of MOS devices has been suggested in this study. The deuterium ion is delivered to the location where the gate oxide resides by using an implantation process and subsequent N2 annealing process at the back-end of metallization process. A conventional MOS field-effect transistor (MOSFET) with a 3-nm-thick gate oxide and poly-to-ploy capacitor sandwiched with 20-nm-thick SiO2 were fabricated in order to demonstrate the deuterium effect in our process. An optimum condition of ion implantation was necessary to account for the topography of the overlaying layers in the device structure and to minimize the physical damage due to the energy of the implanted ion. Device parameter variations, the gate leakage current, and the dielectric breakdown phenomenon were investigated in the deuterium-ion-implanted devices. We found the isotope effect between hydrogen- and deuterium-implanted devices and an improved electrical reliability in the deuterated gate oxide. This implies that deuterium bonds are generated effectively at the Si/SiO2 interface and in the SiO2 bulk.

A MULTI-BEAM AND MULTI-RANGE RADAR WITH FMCW AND DIGITAL BEAM FORMING FOR AUTOMOTIVE APPLICATIONS

In this paper, we propose a multi-beam and multi-range (MBMR) radar with frequency modulated continuous wave (FMCW) waveform and digital beam forming (DBF) algorithm to cover a detection area of long range and narrow angle (150m, §10 - ) as well as short range and wide angle (60m, §30 - ) as a single 24GHz sensor. The developed radar is highly integrated with multiple phased-array antennas, a two-channel transmitter and a four-channel receiver using K-band GaAs RF ICs, and back-end processing board with subspace- based DBF algorithm. The proposed 24GHz MBMR radar can be used for an adaptive cruise control (ACC) stop-and-go system which typically consists of three radars, such as two 24GHz short-range radars for object detection in an adjacent lane and one 77GHz long- range radar for object detection in the center lane.

Stacked-Spiral RF Inductor With Vertical Nano-Powder Magnetic Core in CMOS

This letter reports a novel concept of vertical stacked-spiral RF inductor with integrated nano-powder-magnetic-core in standard CMOS. A proof-of-concept design in a foundry 0.18 6-metal CMOS using a new post-CMOS backend process (CMOS+) is depicted. Measurement shows significant improvement in inductance of 35% and quality factor of 41% up to multi-GHz in prototype. An -density of 420 at 1.2 GHz is achieved. Limiting factors for the relatively low factor in the prototypes and improvement method are discussed.

Low-Loss Amorphous Silicon Multilayer Waveguides Vertically Stacked on Silicon-on-Insulator Substrate

Low-loss amorphous-silicon (a-Si) waveguides comprising three vertically stacked layers prepared on silicon-on-insulator substrates are demonstrated. We have fabricated multilayer a-Si waveguides and investigated their loss characteristics; this is the first such investigation to our knowledge. All the process temperatures were regulated below 400 °C for the complementary metal oxide semiconductor (CMOS) backend process compatibility. When the surface roughness and sidewall roughness were decreased, the propagation loss decreased to 3.7 dB/cm even in the case of the third layer a-Si waveguide. Such low-loss waveguides can be effectively applied to realize multilayer stacked optical devices.

A MIM Diode with Ultra Abrupt Switching Process and High On/Off Current Ratio

In this work, a TaOx-based Metal-Insulator-Metal (MIM) diode is demonstrated. This Cu/TaOx/W diode is fabricated with sputtering method without any high temperature process (~ 400oC), which shows promising compatabiltiy with CMOS back-end process. The fabricated diode exhibits abrupt switching process (~3mV/dec) with small switching voltage (~ 0.7V), high on/off current ratio more than 106, low leakage current (~ pA). Current is limited when the diode is switched to the on-state to avoid the hard breakdown caused by the thermal run-away. On/off current ratio can be further improved to 108 by the limited current increase at the precondition of ensuring the safty of the diode. This MIM diode can be easily integrated into high density PCM or RRAM cross-point arrays due to its simple fabrication process. Its nearly ideal diode behavior with ultra low on-resistance can also be used as an ESD protection device or as a rectifying element in circuit.

HHNEC 0.18um BCD Technology for High Density Power Integration

A new 0.18um Bipolar CMOS DMOS (BCD) integrated platform is presented. This technology provides both N type and P type power LDMOS with capability of voltage up to 40V applications; these power LDMOS devices have competitive specific on-resistance; the parasitical miller capacitance and SOA of power LDMOS devices are optimized by careful device and layout design; it also provides HVCMOS with voltage up to 40V analog application; the bipolar devices in this technology are also optimized for power switching as standard bipolar process offers. The 1.8V/5.0V CMOS devices are performance compatible with industry main stream 1.8V/5.0V CMOS technology. This platform is modularized with various options; it can be separated into 1.8V/5.0V mixed 40V HVCMOS process, 5.0V Vgs 40V Vds 0.18um backend BCD process and 1.8V/5.0V Vgs 40V Vds 0.18um BCD process, which customers can easily choose for their product needs without any process tuning.

Back-end soft and hard defect monitoring using a single test chip

The objective of this paper is to present a mixed test structure designed to characterize yield losses due to hard defect and back-end process variation (PV) at die and wafer level. A brief overview of the structure, designed using a ST-Microelectronics’ 130 nm technology, is given. This structure is based on a SRAM memory array for detecting hard defects. Moreover each memory cell can be configured in the Ring Oscillator (RO) mode for back-end PV characterization. The structure is tested in both modes (SRAM, RO) using a single test flow. The test data analysis method is presented and applied to experimental results to confirm the ability of the structure to monitor PV and defect density.