Heterogeneous Integration Approach Research Articles

SiNx‐Based Digital–Analog Hybrid Resistive Random Access Memory via Heterogeneous Integration

Herein, a digital–analog hybrid resistive random‐access memory (RRAM) is prepared by integrating the structurally similar SiNx‐based digital‐type RRAM with analog‐type RRAM by heterogeneous integration method. The Pt/SiNx/Ta/Ru is a digital RRAM due to the formation and breakage of the internal silicon dangling bonds conductive filaments, and the TiN/SiNx/Ta/Ru structure is an analog RRAM due to the traps‐filled limit region of the space‐charge limited current. The heterogeneously integrated digital RRAM has good cycling stability and endurance characteristics, and the heterogeneously integrated analog RRAM has high linearity and multistate counting characteristics. This heterogeneous integration approach is useful for the implementation of hybrid digital–analog RRAM.

Heterogeneous integration of III–V semiconductor lasers on thin-film lithium niobite platform by wafer bonding

Thin-film lithium niobate (TFLN) photonic integrated circuits (PICs) have emerged as a promising integrated photonics platform for the optical communication, microwave photonics, and sensing applications. In recent years, rapid progress has been made on the development of low-loss TFLN waveguides, high-speed modulators, and various passive components. However, the integration of laser sources on the TFLN photonics platform is still one of the main hurdles in the path toward fully integrated TFLN PICs. Here, we present the heterogeneous integration of InP-based semiconductor lasers on a TFLN PIC. The III–V epitaxial layer stack is adhesively bonded to a TFLN waveguide circuit. In the laser device, the light is coupled from the III–V gain section to the TFLN waveguide via a multi-section spot size converter. A waveguide-coupled output power above 1 mW is achieved for the device operating at room temperature. This heterogeneous integration approach can also be used to realize on-chip photodetectors based on the same epitaxial layer stack and the same process flow, thereby enabling large-volume, low-cost manufacturing of fully integrated III–V-on-lithium niobate systems for next-generation high-capacity communication applications.

Pulsed laser welding of macroscopic 3D graphene materials.

Welding is a key missing manufacturing technique in graphene science. Due to the infusibility and insolubility, reliable welding of macroscopic graphene materials is impossible using current diffusion-bonding methods. This work reports a pulsed laser welding (PLW) strategy allowing for directly and rapidly joining macroscopic 3D porous graphene materials under ambient conditions. Central to the concept is introducing a laser-induced graphene solder converted from a designed unique precursor to promote joining. The solder shows an electrical conductivity of 6700 S m-1 and a mechanical strength of 7.3 MPa, over those of most previously reported porous graphene materials. Additionally, the PLW technique enables the formation of high-quality welded junctions, ensuring the structural integrity of weldments. The welding mechanism is further revealed, and two types of connections exist between solder and base structures, i.e., intermolecular force and covalent bonding. Finally, an array of complex 3D graphene architectures, including lateral heterostructures, Janus structures, and 3D patterned geometries, are fabricated through material joining, highlighting the potential of PLW to be a versatile approach for multi-level assembly and heterogeneous integration. This work brings graphene into the laser welding club and paves the way for the future exploration of the exciting opportunities inherent in material integration and repair.

An ABGE-aided manufacturing knowledge graph construction approach for heterogeneous IIoT data integration

The Industrial Internet of Things (IIoT) provides a foundation for the development of emerging digital servitization paradigm in smart manufacturing. The deep integration of massive heterogeneous IIOT data plays a critical role in realising manufacturing digital servitization. However, there is a knowledge gap between different manufacturing fields, which brings a challenge for efficient integration and leverage of industrial big data. For this purpose, a Framework of Manufacturing Knowledge Graph (FMKG) is proposed, which is used to extracts industry knowledge triples from multi-source heterogeneous data to integrate domain knowledge. Also, an attention-based graph embedding model (ABGE) is proposed to discover and complement the implicit missing relationships in the knowledge graph to obtain a complete industrial knowledge graph. The effectiveness of the ABGE model has been verified on several knowledge graph data sets. And an aerospace enterprise production process was taken as an example to establish a product quality knowledge graph, which proved the feasibility of the proposed method.

Monolithic InGaN Multicolor Light‐Emitting Devices

Given the recent surge of interest in full‐color microLED display technologies, researchers in academia and industry have been looking for feasible and cost‐effective solutions to realize multicolor emission from a single wafer to overcome the shortcomings of existing “mass transfer” approaches that involve the assembly of huge numbers of chips from multiple wafers. In contrast to such heterogeneous integration approaches, monolithic integration solutions that combine different color emitters onto a single chip or wafer offer potentially higher yield, scalability, robustness, efficiency, and manufacturability. In this review, an overview of the main monolithic integration approaches for the assembly of polychromatic emitters on a chip or wafer, including top‐down fabrication and bottom‐up growth approaches is given. The successful realization of monolithic polychromatic emission would greatly speed up the development of phosphor‐free white light sources, chip‐scale color microLED displays, and other GaN‐based optoelectronic systems and platforms.

Silicon-integrated nonlinear III-V photonics

Mainstream silicon photonic integrated circuits are based on compact and low-loss silicon-on-insulator (SOI) waveguide platforms. However, monolithic SOI-based photonics provides only a limited number of functional device types. Here, to extend the on-chip capabilities, we propose a general heterogeneous integration approach to embed highly nonlinear III-V (AlGaAs) photonics into the SOI platform. We develop low-loss AlGaAs-on-SOI photonic circuits with integrated Si waveguides and showcase sub-milliwatt-threshold ( ∼ 0.25 mW ) Kerr frequency comb generation in ultrahigh- Q AlGaAs microrings ( Q over 10 6 ) at the telecom bands. Our demonstration complements existing mature Si photonics technology with efficient nonlinear functionalities provided by III-V and propels conventional Si photonics into emerging nonlinear photonic applications towards fully chip-based nonlinear engines.

High-density low-loss millimeter-wave package interconnects with the impact of dielectric-material surface roughness

Heterogeneous package integration and chiplet approaches are the key technology to enable next-generation high performance small form-factor packages for emerging applications. Millimeter-wave packaging for fifth-generation and upcoming sixth-generation platforms also need to meet the high-density low signal-loss interconnect specifications utilizing advanced conductor and dielectric materials. This article presents the comparison of the liquid-based photoimageable dielectric (PID) and dry-film dielectric materials in terms of interconnect path losses that are critical in mm-wave frequency bands. The conductor loss being more dominant in the frequency bands and in thinner dielectric structures, we assess daisy chains and microstrip lines on 15-μm dielectric by measuring the S-parameters to quantify the impact of the surface roughness at around 28 GHz. Measured results from the daisy chain and microstrip line structures exhibit that the smooth surface of the liquid-based PID (3 nm) leads to 8%–32% lower signal loss in the dB scale than the 325-nm rough dry-film dielectric. The study provides comprehensive experimental results that the different material forms with various surface roughness largely impact the package-level interconnect loss.

Digital Electrochemistry: Digital Electrochemistry for On‐Chip Heterogeneous Material Integration (Adv. Mater. 26/2021)

In article number 2101272, Bin Bao, Daniil Karnaushenko, Oliver G. Schmidt, and co-workers develop a digital electrochemistry approach for on-chip heterogeneous material integration. An amorphous indium-gallium-zinc oxide thin-film-transistor (a-IGZO TFT) active-matrix is used to address pixels within the matrix and locally control electrochemical reactions for material growth and actuation. Multicolored electrochromic patterns and actuator arrays can be fabricated. The approach can be extended to a broad range of materials and structures, opening up a new path for advanced heterogeneous microsystem integration.

Straining and Tuning Atomic Layer Nanoelectromechanical Resonators via Comb‐Drive MEMS Actuators

AbstractBroad frequency tuning is an essential attribute desired in resonant nano/microelectromechanical systems (NEMS/MEMS) and their many applications. Endowed with ultrahigh intrinsic strain limits, combined with other unconventional properties, atomically thin 2D crystalline materials are excellent candidates for building highly tunable resonant NEMS. Here a heterogeneous integration approach is demonstrated to enable on‐chip, continuous, and broad frequency tuning in 2D NEMS resonators by directly controlling strain via voltage‐controlled silicon‐on‐insulator (SOI) comb‐drive MEMS actuators. By varying the comb‐drive actuation voltage, resonance frequency of the 2D NEMS can be tuned as large as 75% continuously with precise control. The comb‐drive actuation‐enabled direct straining and tuning also yield quality (Q) factor boost up to twofold. It is validated that this technique is readily applicable to straining and tuning representative 2D NEMS in various leading materials: graphene, molybdenum disulfide (MoS2), and hexagonal boron nitride (h‐BN). This study demonstrates that additively integrating 2D resonators atop mainstream SOI MEMS enables a versatile platform, and opens new possibilities for voltage control and broad tuning of 2D NEMS on chip.

Multiwavelength membrane laser array using selective area growth on directly bonded InP on SiO2/Si

The cost and power consumption of optical transmitters are now hampering further increases in total transmission capacity within and between data centers. Photonic integrated circuits (PICs) based on silicon (Si) photonics with wavelength-division multiplexing (WDM) technologies are promising solutions. However, due to the inefficient light emission characteristics of Si, incorporating III-V compound semiconductor lasers into PICs via a heterogeneous integration scheme is desirable. In addition, optimizing the bandgap of the III-V material used for each laser in a WDM transmitter becomes important because of recent strict requirements for optical transmitters in terms of data speed and operating temperature. Given these circumstances, applying a direct-bonding scheme is very difficult because it requires multiple bonding steps to bond different-bandgap III-V materials that are individually grown on different wafers. Here, to achieve wideband WDM operation with a single wafer, we employ a selective area growth technique that allows us to control the bandgap of multi-quantum wells (MQWs) on a thin InP layer directly bonded to silicon (InP-on-insulator). The InP-on-insulator platform allows for epitaxial growth without the fundamental difficulties associated with lattice mismatch or antiphase boundaries. High crystal quality is achieved by keeping the total III-V layer thickness less than the critical thickness (430 nm) and compensating for the thermally induced strain in the MQWs. By carrying out one selective MQW growth, we successfully fabricated an eight-channel directly modulated membrane laser array with lasing wavelengths ranging from 1272.3 to 1310.5 nm. The fabricated lasers were directly modulated at 56-Gbit/s with pulse amplitude modulation with four-amplitude-level signal. This heterogeneous integration approach paves the way to establishing III-V/Si WDM-PICs for future data-center networks.

Heterogeneous III-V on silicon nitride amplifiers and lasers via microtransfer printing

The development of ultralow-loss silicon-nitride-based waveguide platforms has enabled the realization of integrated optical filters with unprecedented performance. Such passive circuits, when combined with phase modulators and low-noise lasers, have the potential to improve the current state of the art of the most critical components in coherent communications, beam steering, and microwave photonics applications. However, the large refractive index difference between silicon nitride and common III-V gain materials in the telecom wavelength range hampers the integration of electrically pumped III-V semiconductor lasers on a silicon nitride waveguide chip. Here, we present an approach to overcome this refractive index mismatch by using an intermediate layer of hydrogenated amorphous silicon, followed by the microtransfer printing of a prefabricated III-V semiconductor optical amplifier. Following this approach, we demonstrate a heterogeneously integrated semiconductor optical amplifier on a silicon nitride waveguide circuit with up to 14 dB gain and a saturation power of 8 mW. We further demonstrate a heterogeneously integrated ring laser on a silicon nitride circuit operating around 1550 nm. This heterogeneous integration approach would not be limited to silicon-nitride-based platforms: it can be used advantageously for any waveguide platform with low-refractive-index waveguide materials such as lithium niobate.

Recent advances of heterogeneously integrated III–V laser on Si

Due to the indirect bandgap nature, the widely used silicon CMOS is very inefficient at light emitting. The integration of silicon lasers is deemed as the ‘Mount Everest’ for the full take-up of Si photonics. The major challenge has been the materials dissimilarity caused impaired device performance. We present a brief overview of the recent advances of integrated III–V laser on Si. We will then focus on the heterogeneous direct/adhesive bonding enabling methods and associated light coupling structures. A selected review of recent representative novel heterogeneously integrated Si lasers for emerging applications like spectroscopy, sensing, metrology and microwave photonics will be presented, including DFB laser array, ultra-dense comb lasers and nanolasers. Finally, the challenges and opportunities of heterogeneous integration approach are discussed.

Novel Scalable Transfer Approach for Discrete III‐Nitride Devices Using Wafer‐Scale Patterned h‐BN/Sapphire Substrate for Pick‐and‐Place Applications

AbstractThe mechanical release of III‐nitride devices using h‐BN is a promising approach for heterogeneous integration. Upscaling this technology for industrial level requires solutions that allow a simple pick‐and‐place technique of selected devices for integration while preserving device performance. An advance that satisfies both of these requirements is demonstrated in this work. It is based on a lateral control of the h‐BN quality, using patterned sapphire with a SiO2 mask, to achieve localized van der Waals epitaxy of high‐quality GaN based device structures. After process fabrication, the devices can be individually picked and placed on a foreign substrate without the need for a dicing step. In addition, this approach could reduce delamination of h‐BN on large diameter substrates because each h‐BN region is smaller, with independent device structures. Discrete InGaN LEDs on h‐BN are grown and fabricated on 2 in. patterned sapphire using a SiO2 mask. A set of devices are selectively released and transferred to flexible aluminum tape. The transferred LEDs exhibit blue light emission around 435 nm. The approach presented here is scalable on any wafer size, can be applied to other types of nitride‐based devices, and can be compatible with commercial pick‐and‐place handlers for mass production.

Combining gene expression data and prior knowledge for inferring gene regulatory networks via Bayesian networks using structural restrictions.

Gene Regulatory Networks (GRNs) are known as the most adequate instrument to provide a clear insight and understanding of the cellular systems. One of the most successful techniques to reconstruct GRNs using gene expression data is Bayesian networks (BN) which have proven to be an ideal approach for heterogeneous data integration in the learning process. Nevertheless, the incorporation of prior knowledge has been achieved by using prior beliefs or by using networks as a starting point in the search process. In this work, the utilization of different kinds of structural restrictions within algorithms for learning BNs from gene expression data is considered. These restrictions will codify prior knowledge, in such a way that a BN should satisfy them. Therefore, one aim of this work is to make a detailed review on the use of prior knowledge and gene expression data to inferring GRNs from BNs, but the major purpose in this paper is to research whether the structural learning algorithms for BNs from expression data can achieve better outcomes exploiting this prior knowledge with the use of structural restrictions. In the experimental study, it is shown that this new way to incorporate prior knowledge leads us to achieve better reverse-engineered networks.

Intelligent Approach for Heterogeneous Data Integration: Information Processes Analysis Engine in Clinical Remote Monitoring Systems

The paper presents a research project which aimed to design and develop an intelligent approach for the integration of heterogeneous data and knowledge sources in personalized healthcare. The integration profile we propose is mainly focused on remote monitoring of patients with arterial hypertension and chronic heart failure. This clarifies main temporal data used by event detection systems and limits it to systolic/diastolic pressure and heart rate. Our aim is to design and propose software component for such a system which makes it possible to reconstruct the whole remote monitoring and treatment process by connecting event logs. The analysis of the reconstructed process done by comparing with an expert system which contains different rules for monitoring/treatment process optimization.

Heterogeneous Integration of Three-Primary-Color Photoluminescent Nanoparticle Arrays with Defined Interfaces.

Minimized photoluminescent devices require both high-density fluorescent arrays and minimal cross-talk between neighboring pixels on the limited area. However, the challenges to achieve the overall integration of nanomaterial-based devices hinder the development of microscale full-color displays, including micro/nanoarray density, orientation control, multimaterial interface morphology, and uniform colors. Here, we report a heterogeneous integration approach to control the orientation, combination, and density of fluorescent micro/nanoarrays on flexible substrates. By controlling the defined interface and critical shrinkage width of liquid bridges, the width of three-primary-color micro/nanolines reached 100 nm. The interval between two parallel luminous lines is down to 40 μm, and the optical cross-talk effect is remarkably reduced. This work provides a facile approach to prepare high-performance micro-photoluminescent and imaging arrays for next-generation flexible display and lighting technology.

(Invited) Adhesive Wafer Bonding for Heterogeneous System Integration

Heterogeneous integration of micro- and nano-electromechanical systems (MEMS, NEMS) and photonic devices on top of integrated circuits (ICs) is an emerging technology that will lead to highly compact devices combining traditional ICs with advanced sensors and actuators. Different components, such as MEMS, photonic devices, and complementary metal oxide semiconductor (CMOS) circuits are fabricated on separate substrates and subsequently joined into one single substrate by wafer bonding [1,2]. This enables the integration of high-performance MEMS, NEMS, and photonic devices made of materials such as mono-crystalline silicon, germanium, III–V materials, piezoelectric materials, shape memory alloys (SMAs), carbon nanotubes (CNT), or nanowires directly on top of ICs, which is otherwise not possible. We review heterogeneous integration approaches for various MEMS, NEMS, and photonics systems based on layer transfer by adhesive wafer bonding. The main advantages of adhesive wafer bonding are the low bonding temperatures (20 - 450°C), its insensitivity to wafer surface topology, the compatibility with standard CMOS wafers, and the ability to join practically any wafer materials [3]. Adhesive wafer bonding does not require special wafer surface treatments, such as planarization or excessive cleaning, and therefore, offers a simple and cost-effective transfer process. Layer transfer by adhesive wafer bonding consists of several steps. First, the donor wafer, containing the layer to be transferred, is aligned to the target wafer (e.g. CMOS wafer) and bonded using an intermediate polymer adhesive. Second, the bulk material (e.g. Si substrate) of the donor wafer is sacrificially etched while the transferred layer remains on the target wafer of top of the polymer layer. The transferred layer can subsequently be further processed using conventional process steps to create the desired devices. Compatible processes include structuring and etching of the transferred layer, material deposition such as chemical and physical vapour deposition (CVD, PVD), ion implantation, and annealing amongst others. These processes can be used to create electrical, mechanical, and photonic devices and electrical interconnections between the transferred layers and the underlying substrate (vias). Using adhesive bonding with an intermediate polymer further offers highly selective and controlled sacrificial release etching capabilities in O2 plasma. This provides a simple approach for realizing free-standing MEMS and NEMS structures without wet etching and critical point drying (CPD), thus increasing manufacturing yield and reducing fabrication complexity. Lastly, the presented heterogeneous integration approach is compatible with pre-structuring of the target substrates, as well as of the transfer layer. Thus, it is possible to integrate not only layers of high-performance materials, but pre-processed devices and systems. We illustrate the versatility of adhesive wafer bonding for heterogeneous integration with different application examples of MEMS and optical systems integrated onto various target substrates. Each example requires a slightly different heterogeneous integration approach. A suspended Si/SiGe quantum-well infrared microbolometer array was integrated on top of a CMOS substrate by adhesive transfer bonding of the Si/SiGe thermistor layer and subsequent patterning of the bolometer array [4]. Since the bolometer pixels needed to be suspended, Ni pillars were formed and the polymer adhesive was sacrificially removed to free-etch the bolometer pixels. The Ni pillars provide both electrical connection to the underlying CMOS circuits and mechanical support for the suspended bolometer pixels. For the integration of nano-electromechanical (NEM) switches for ultra-low power computing applications, TiW vias were used as electrical interconnects. In order to achieve short signal path lengths and thereby increase the performance and energy efficiency of the circuits, an ultra-thin adhesive bonding layer of just 200 nm was used. In optical systems that don’t require electrical interconnections, the polymer adhesive itself can be used as mechanical support. Finally, the capability for multiple layer transfer steps was demonstrated by fabricating hidden hinge micromirror arrays [5]. In this approach, a layer of mono-crystalline silicon was transferred onto the target wafer and the mirror hinges were subsequently patterned in this layer. Pillars extending from the hinge layer to the target substrate were formed, but the polymer adhesive was not removed at this stage. Instead, a second mono-crystalline silicon layer was transferred on top of the hinge layer using adhesive bonding. The mirrors are then formed in this second transferred layer. Pillars connecting the mirrors to the hinges were formed and finally both polymer adhesive layers were sacrificially removed in O2-plasma, thus, creating free-standing micromirrors. In conclusion, the flexibility to precisely control the thickness of the adhesive polymer layers and selectively remove it makes adhesive wafer bonding a highly versatile approach for heterogeneous integration of a large variety of MEMS and photonic applications. Figure 1

A two-tiered unsupervised clustering approach for drug repositioning through heterogeneous data integration

BackgroundDrug repositioning is the process of identifying new uses for existing drugs. Computational drug repositioning methods can reduce the time, costs and risks of drug development by automating the analysis of the relationships in pharmacology networks. Pharmacology networks are large and heterogeneous. Clustering drugs into small groups can simplify large pharmacology networks, these subgroups can also be used as a starting point for repositioning drugs. In this paper, we propose a two-tiered drug-centric unsupervised clustering approach for drug repositioning, integrating heterogeneous drug data profiles: drug-chemical, drug-disease, drug-gene, drug-protein and drug-side effect relationships.ResultsThe proposed drug repositioning approach is threefold; (i) clustering drugs based on their homogeneous profiles using the Growing Self Organizing Map (GSOM); (ii) clustering drugs based on drug-drug relation matrices based on the previous step, considering three state-of-the-art graph clustering methods; and (iii) inferring drug repositioning candidates and assigning a confidence value for each identified candidate. In this paper, we compare our two-tiered clustering approach against two existing heterogeneous data integration approaches with reference to the Anatomical Therapeutic Chemical (ATC) classification, using GSOM. Our approach yields Normalized Mutual Information (NMI) and Standardized Mutual Information (SMI) of 0.66 and 36.11, respectively, while the two existing methods yield NMI of 0.60 and 0.64 and SMI of 22.26 and 33.59. Moreover, the two existing approaches failed to produce useful cluster separations when using graph clustering algorithms while our approach is able to identify useful clusters for drug repositioning. Furthermore, we provide clinical evidence for four predicted results (Chlorthalidone, Indomethacin, Metformin and Thioridazine) to support that our proposed approach can be reliably used to infer ATC code and drug repositioning.ConclusionThe proposed two-tiered unsupervised clustering approach is suitable for drug clustering and enables heterogeneous data integration. It also enables identifying reliable repositioning drug candidates with reference to ATC therapeutic classification. The repositioning drug candidates identified consistently by multiple clustering algorithms and with high confidence have a higher possibility of being effective repositioning candidates.

Ontology based heterogeneous materials database integration and semantic query

Materials digital data, high throughput experiments and high throughput computations are regarded as three key pillars of materials genome initiatives. With the fast growth of materials data, the integration and sharing of data is very urgent, that has gradually become a hot topic of materials informatics. Due to the lack of semantic description, it is difficult to integrate data deeply in semantic level when adopting the conventional heterogeneous database integration approaches such as federal database or data warehouse. In this paper, a semantic integration method is proposed to create the semantic ontology by extracting the database schema semi-automatically. Other heterogeneous databases are integrated to the ontology by means of relational algebra and the rooted graph. Based on integrated ontology, semantic query can be done using SPARQL. During the experiments, two world famous First Principle Computational databases, OQMD and Materials Project are used as the integration targets, which show the availability and effectiveness of our method.

Thermal Management of Monolithic Versus Heterogeneous Lasers Integrated on Silicon

Thermal management is one of the most important issues for today's microelectronics design. Photonics is expected to bring significant advantages to future beyond-CMOS microprocessors in terms of heat dissipation with the use of optical interconnects but thermal management of integrated laser sources remains an issue. In this paper, we investigate the thermal properties of various III-V semiconductor laser sources integrated on silicon. Monolithic (through GaSb epitaxy) and heterogeneous (through InP/SOI bonding) integration approaches are studied with both ridge and microresonators laser geometries. It is first shown that the presence of dislocations in monolithically integrated structures, as determined by transmission electron microscopy, have no impact on thermal conductivities. Thermal mappings of the devices are then computed. It is demonstrated that heterogeneous lasers are thermally more insulated from the surrounding CMOS than monolithic lasers (especially at the back side), which avoids photonic-induced hot spots formation on the chip, but can degrade the overall laser performances.