Embedded Flexible Hybrid Electronics for the Internet of Things

After the World Wide Web and the Mobile Internet, we are now heading to the third and potentially most “disruptive” phase of the Internet revolution – the “Internet of Things” (IoT), defined as the worldwide network of interconnected objects uniquely addressable based on standard communication protocols [1]. The original IoT definition evolved to include three interrelated and interconnected paradigms—internet-oriented (network of smart objects), things-oriented (sensors), and semantic-oriented (knowledge). The usefulness of IoT can be unleashed only in an application domain where the three paradigms intersect [2–4]. Ashton points out that the IoT has the potential to change the world, just as the Internet did [1]. The U.S. National Intelligence Council (NIC) included IoT among the list of six disruptive civil technologies with a potential significant impact on US strategic interests [5]. NIC anticipates that “by 2025 Internet nodes may reside in everyday things—food packages, furniture, paper documents, and more” [5].

The IoT objects, or “things”, fall in several broad categories as shown in Fig. 1: (a) passive objects with very limited identification features based on the Radio Frequency Identification (RFID) technology, (b) passive RFID objects with extra memory and sensing capabilities, and (c) active, or smart, objects augmented with sensing, processing, and network capabilities based on non-RFID communication protocols. In fact, some definitions of IoT are entirely based on the latest category: smart objects organized as a loosely-coupled, decentralized system of autonomous physical/digital objects [6, 7]. In contrast with the passive objects, these smart objects can make sense of their environment and can interact with human users and other nearby smart objects [6]. These smart object are undoubtedly in the heart of the IoT but they alone cannot realize the full potential of IoT.

To be considered “citizens” of the IoT ecosystem, the IoT things must possess networking capabilities. Wireless technologies such as Bluetooth, Wi-Fi, ZigBee, etc. can and are utilized for that purpose. However, the most ubiquitous objects in the IoT ecosystem, those that reside at the lowest system level, are based on the passive (no on-board power) RFID technology. RFID is the enabling technology for the IoT objects as it encompasses three critical functions: object identification, data acquisition (if coupled with a sensor) and storage, and communication capabilities, all delivered in a very low-cost package. As such, the passive RFID technology is expected to play a key role in the IoT [2, 8]. This fact is recognized in the concept of the Unified Internet of Things Architecture [9], such as this depicted in Fig.1, where IoT integrates both the RFID and smart object-based infrastructures. In this concept, RFID objects support fundamental data acquisition functionalities, while smart objects assume more complex functionalities. In the form of wireless passive sensors, the RFID objects are found in packaging, form the backbone of the structural health monitoring systems, provide non-invasive and continuous monitoring of physiological parameters, etc.

The IoT as a technology is a step behind the IoT as a concept. Here, IoT technology is defined in a narrow sense as the practical application of engineering knowledge for creation of IoT objects. The easiest and simplest way to do that is just to add IoT functionally to an item. RFID capability is already added to everyday items in the physical form of adhesive “smart” labels, enabling them to become parts of the IoT ecosystem. This “add-on” legacy approach is in essence an adaptation of the existing label technology intended to serve the needs of a new application. As such it fails to deliver the attributes required for ubiquitous IoT applications: low cost, unobtrusiveness, security, reliability. Therefore, a change in the technology paradigm is needed to address the unique aspects of the IoT vision. New objects should be designed as native IoT items where the wireless connectivity is not added but integrated in the object. This is already done for physically large and structurally complex items such as smart homes, vehicles, home appliances, transportation and agriculture equipment, security cameras and systems, consumer electronics, etc. However, very little has been accomplished when it comes to integrating IoT functionality in most of the simple objects surrounding us.

It can be argued that the complete, all-inclusive, implementation of the IoT concept will in part depend on the ability of the industry to produce intelligent item-level packaging, defined here as packaging which has the ability to sense or measure an attribute of the product. Let's consider, for example, a scenario where an ordinary pack of ground meat is placed in a refrigerator. In the IoT realm, the meat (a passive IoT object) should be able to communicate its expiration date and freshness to the refrigerator (a smart IoT object), which, in turn, communicates this information to the human user. If this scenario is to be realized today, a standalone, disparate wireless RFID sensor would be added to the package. It is obvious that this “add-on” approach is not reliable, its application increases the implementation cost and, most probably, will raise bio- and food compatibility concerns. An intelligent package, where the sensor is seamlessly and unobtrusively integrated in the packaging material, will avoid the shortcoming of the add-on solution.

In other examples, the add-on electronic devices may impact negatively the host item's form factor and appearance. The add-on devices that rely on laminating multiple layers of material with a chip sandwiched inside are also counterfeit- and tampering-prone as the device, which is also used to uniquely identify/authenticate the product, can be easily removed from one item and after tampering re-laminated on another. Embedded electronic solutions in the native IoT items minimize these vulnerabilities. Since the electronics are embedded within an individual layer of material, getting at it with-out destroying the host material is extremely difficult. Replacing a chip with an alternate and leaving behind no evidence of tampering is even more difficult.

The RFID IoT objects are predicted to be produced in volumes measured in hundreds of billions. DaCosta stressed that the key economic factor that will drive the deployment of the IoT is the cost at the end points [10]. Obviously, the future of the IoT depends on developing an ultra-low-cost technology solution that can mass-produce low-cost sensing RFID IoT objects on flexible substrates. A common assumption today is that all-printed electronics is the enabling technology that will propel a rapid growth in smart and intelligent packaging [11], and, from there, will enable the IoT implementation [12–14]. Scientists and engineers all over the world are developing printed organic circuits working on the promise that novel RFID and sensor devices will be printed like today's newspaper at high speed, in large amounts, and at a very low cost. Over the last decade, there has been a significant increase in the efforts dedicated to the development and implementation of organic electronic components on flexible substrates, including display and lightning, solar cell, battery and passive electronic components (conductors, antennas, resistors, etc.). Research groups have demonstrated other types of application, such as sensing [15–17], critical for the IoT applications.

The proponents of all-printed electronics almost unanimously select the RFID tag as a technology demonstration platform because RFID tags are simple enough devices that must be extremely cheap, flexible, and are produced in huge volumes, all of these characteristics being the tenets of the all-printed electronics paradigm. It is not a stretch to extrapolate this line of reasoning to the sensing RFID objects. However, printed electronics hits a roadblock when attempts are made to print high-performance semiconductor devices such as fully-functional ICs. There is a growing consensus among researchers and practitioners that all-printed circuits of any significant functionality are still far in the future. This has given rise to the concept of flexible hybrid electronics (FHE). The U.S. Government has recognized the strategic importance of FHE releasing a solicitation for proposals to initiate and sustain a Flexible Hybrid Electronics Manufacturing Innovation Institute (FHE-MII) as part of the National Network for Manufacturing Innovation (NNMI) [18]. IoT is identified in this program as one of the major FHE areas of application. It is noteworthy to mention that the printed technology for active electronic components such as ICs is explicitly excluded from this program as it is considered by the Government to be still in the early stages of development.

FHE can be simply described as flexible, thinned silicon CMOS chips and other discrete components packaged on a conformable printed substrate. Integrating printed and CMOS electronics draws benefits from both worlds. The functions and performance necessary in data processing and modern communication delivered by the conventional CMOS technology are integrated in the cost-effective printed electronics technology with the resulting hybrid systems having potential for various low-cost, disposable electronic applications, incl. sensing RFID IoT objects. In a FHE RFID-based sensor, the methods of printed electronics are used to produce passives, interconnects, antennas, sensing elements, etc., using the conventional printing techniques such as screen printing, ink jet, gravure, flexography and others. Independently, discrete components such as ICs are readied for assembly. The preparation includes thinning the CMOS device to a thickness of less than 50 μm, at which thickness the inherently brittle Si becomes conformal to the flexible substrate. During the packaging step, the thinned ICs are attached to the circuitry on the flexible printed substrate. The completed device is then added to an item or, in case of intelligent packaging, the device is integrated in the packaging material (paper, cardboard, plastic sheet) during the material's fabrication process. The latter case is discussed more in detail in this text.

A sensing RFID IoT object is comprised of a sensing element(s) and a small microchip attached to an antenna. It is based on the passive RFID concept meaning it does not have an onboard power supply. It harvests the energy required for transmitting from the interrogating signal provided by a RFID reader. From a physical point of view the RFID-enabled IoT object needs to be flexible and miniature. It needs to be miniature to accommodate a wide variety of form factors. It needs to be flexible so that it can conform to all types of surface geometries to which it is attached. If it is intended for integration, it needs also to be ultrathin to fit inside a sheet of ordinary materials such as paper and plastics.

To achieve the desired flexibility and thickness, the chip in the RFID-enabled IoT object must be less than 50 μm thick, preferably less than 30 μm. Such ultrathin chips have been around for years. For example, more than ten years ago Hitachi demonstrated a series of ultrathin silicon-on-insulator (SOI)-based RFID chips called a “μ-chip,” including a 75 μm/side, 7.5 μm thick chip [19–22]. Ultrathin SOI chips have been also demonstrated by other groups [23–26]. SOI is a technology well-suited for ultrathin Si chips but it is expensive and not readily available [27]. Alternatively, the ultrathin bulk Si technology has been and is still being actively investigated, mostly in Europe [28].

Burghartz and co-authors provided an in-depth discussion on the ultrathin chip technology and applications in a recent publication [29]. However, very little has been reported in the area of ultrathin chip assembly. Various methods have been proposed with the majority of the work done using prototype equipment redesigned for handling thin silicon, mostly flip-chip die bonders with adapted tooling and special release tapes [30–33]. Detailed discussion on this methods is available elsewhere [34–36] but the overall conclusion is that the conventional pick-and-place (PnP) tools cannot sufficiently handle chips with thicknesses of 50 μm or less. Furthermore, there is no PnP die bonding equipment that can handle a chip as thin as 20–25 μm or less [37].

Recently, Uniqarta has developed and successfully demonstrated a modified PnP process for assembly of ultrathin chips using temporary carriers, called Handle-Assisted Packaging (H-AP). Similar techniques have been reported by other research groups [38–40]. The principles of H-AP used for flip-chip assembly of an ultrathin die using anisotropic conductive paste (ACP) are illustrated in Fig. 2. The process involves adhering a thinned wafer to a handle substrate using a temporary bonding material, singulating the wafer/handle stack, and assembling the ultrathin chips to a substrate using a conventional die bonder. The process is completed by removing the backside handle, leaving only the attached die at the assembly site, as seen in Fig. 3 where a 25 μm thick RFID die is ACP-attached to the 7–8 μm thick silver epoxy traces printed on 50 μm thick PET substrate.

In high volume applications, the H-AP process can be also carried out on an R2R assembly line where adhesive dispensing, die placement, adhesive cure and handle removal are each carried out at different locations along the line. As an example, Fig. 4 shows a 25 μm thick RFID die assembled using a high rate assembly line to the 12 μm thick aluminum pads of an RFID antenna on a 75 μm thick paper substrate.

The H-AP process is designed for assembly of ultrathin dies onto various substrates. It can be applied for assembly of dies with various thicknesses; however, H-AP is most efficient in the range of 15–50 μm, where the conventional pick-and-place tools cannot operate efficiently, especially at high rates.

To overcome the problems with PnP handling of ultrathin chips, unique methods such as fluidic self-assembly (FSA) [41, 42] and even single bead-manipulating apparatus using micro vacuum tweezers [19] and others [38, 43] have been suggested as alternatives with various degrees of success. Using the energy of light for transferring semiconductor dies has been reported by Karlitskaya [44, 45], A. Piqué [46–50], J. Sheats [51–54] and J. Rodgers's group [55]. The laser has the potential to place chips with transfer rates an order of magnitude higher than those achievable with the conventional PnP [56, 57]. However, the current methods suffer from significant drawbacks including low precision and placement accuracy. To address these problems, a team from the North Dakota State University, Fargo, ND has developed and demonstrated a process called Laser-Enabled Advanced Packaging (LEAP) [34–36, 58]. A central part of the LEAP technology is the thermo-mechanical Selective Laser-Assisted Die Transfer (tmSLADT) technique designed to overcome the problems with the previously reported laser-assisted die transfer methods.

The basic concept of this method, illustrated in Fig. 5, includes the use of a dual dynamic release layer (DRL) to attach the die to be transferred to a laser-transparent glass carrier. The DRL comprises both a blistering layer and an adhesive layer. tmSLADT does not rely on the kinetic energy of a plume of vaporized material or solely on the gravitational force to transfer the dice, both mechanism used by the other groups. Instead, the laser pulse creates a blister in the blistering layer, thus confining the vaporized material within the DRL without rupturing it. The force exerted by the blister, in addition to the gravitational force of the die, initiates the transfer over the gap. This technology was used to transfer ultrathin (20-μm thick) chips as small as 250 μm per side (Fig. 6a). In 2012, tmSLADT was used to fabricate and demonstrate for the first time a functional electronic device (an RFID tag) packaged using lasers (Fig. 6c).

Integrating an electronic device into paper and other sheet materials can be done in a variety of ways briefly explained in the following sections.

In the dry lamination approach, known in the industry as “conversion,” one or two sheets of paper or plastic are added to the RFID inlay using adhesives. This process is commonly used in the industry today to fabricate electronic tickets, access cards, RFID labels and tags, etc. The problems with this approach are discussed in the previous sections in the context of the add-on approach.

The electronic device can be embedded during the process of making a single ply of paper in a manner similar to this used to embed security threads in banknotes. The ultrathin electronic device packaged on an intermediate paper substrate is inserted into the pulp slurry right after the slurry is spread onto the moving “wire” or screen below, which vibrates to induce micro-turbulence in the stock. At this point, the fibers in the pulp consolidate around the electronic device as the water in the slurry drains through the wire. The wet sheet of paper is then pressed to further remove water using a series of heated drier rolls. To achieve a good quality surface finish, the sheet is rolled through a set of high-pressure rolls (calender stacks). Calendering compacts the viscoelastic paper material and changes its surface properties through pressure, friction, and heat. This process creates a complex set of stress conditions to which the brittle chip material is subjected. Preliminary experiments have indicated that ultrathin FHE devices can survive the paper making process, including the calendering step.

Examples of RFID inlays embedded in a single ply of paper are shown in Fig. 7. Fig. 8 shows examples of how a similar approach can be used to demonstrate embedding an ultrathin RFID inlay in a paper slurry in the paper molding process used to fabricate, for example, egg cartons and similar paper products where calendering is not used.

Experiments were also carried out for embedding ultrathin electronics in polymer sheets. The manufacturing process for these materials is based on the extrusion of a polymer melt and is completely different from the methods used to fabricate paper but, nevertheless, it still lends itself well to singleply embedding, as shown in Fig. 9.

Wet lamination is a process where the electronic device is embedded in the paper between two or more wet sheets of paper. Examples of RFID inlays embedded in paper using this method are shown in Fig. 10.

The embedding process works well when the substrate and host materials are the same or similar, for example, a paper inlay in a paper sheet. In this case, the adhesion between the RFID inlay and the paper sheet is very strong and doesn't allow the removal of the inlay without destroying it. Fig. 11 shows the results of adhesive testing done following the recommendations of ISO 2409 since a test method for evaluating the adhesion of paper to a strip of embedded material is not currently available. ISO 2409 describes a test method for assessing the resistance of paint coatings to separation from substrates when a right-angle lattice pattern is cut into the coating, penetrating through to the substrate. The pattern is then brushed lightly with a soft brush several times back and forth along each of the diagonal lines of the lattice pattern, after which the pattern is examined for flaking. As seen in Fig. 11, non or minimal flaking was observed after brushing in all test sites where a lattice pattern was cut in the host material in areas located above the embedded inlays.

In case of dissimilar material, for example PET inlays embedded in paper, adhesives such as polyvinyl acetate-based adhesives or similar should be used to promote the adhesion between the inlay and the host material.

IoT is a disruptive technology which has the potential to change the world by creating information highways in the realm of physical objects. The passive RFID sensor technology is expected to play a key role in this new paradigm. The current, legacy, approach in which the IoT functionally is simply added to an item to make it a part of the IoT ecosystem fails to deliver the attributes required for ubiquitous IoT applications: low cost, unobtrusiveness, security, reliability. New objects should be designed and built as native IoT items where the wireless connectivity is not added but integrated in the object. Such objects will be produced in volumes measured in billions and even trillions. This will be economically feasible only if the industry develops an ultra-low-cost technology solution that can mass-produce sensing RFID-enabled IoT objects on flexible substrates. The advancement of the Flexible Hybrid Electronics (FHE) concept, which combines the methods of printed electronics for passive components with the functionality of thinned flexible CMOS integrated circuits, holds the promise of delivering this capability. A critical step in this technology are the processes for thinning and, especially, for attaching ultrathin chips to flexible substrates at high production rates and low cost. The conventional pick-and-place tools cannot sufficiently handle such chips. The two methods for ultrathin chip assembly presented in this text overcome the limitations of the traditional technology. The Handle-Assisted Packaging (H-AP) method is capable of attaching chips with a thickness as small as 15–20 μm using conventional PnP tools. The Laser-Enabled Advanced Packaging (LEAP) technology is capable of assembling not only ultrathin but also ultra-small chips with lateral dimensions below the limits of the current PnP tools. Both methods use COTS wafers and a series of conventional operations to prepare the ultrathin chips for assembly.

The concept of ubiquitous native IoT objects requires embedding the electronic devices in thin ordinary materials such as paper and plastic sheets. This concept requires new methods where the device is integrated in the sheet material very early in the manufacturing process of this material. Methods for a single- and multiply-ply integration and lamination result in an electronic device that becomes an integral part of the sheet material and as such can be used for IoT applications such as smart packaging, financial and security documents, unobtrusive sensors, and similar.

Laser Enabled Advanced Packaging (LEAP): The author thanks his colleagues O. Swenson, N. Schneck, R. Miller, F. Sarwar, M. Semler, J. Yan, Zh. Chen, and S. Datta, who through the years provided expertise that greatly assisted the research in various capacities. The support from the staff of the North Dakota State University's Center for Nanoscale Science and Engineering is gratefully acknowledged.

The work in this area was supported in part by the Defense Microelectronics Activity (DMEA) under agreements numbers H94003-08-2-0805, H94003-09-2-0905, and H94003-11-2-1102. The U.S. Government is authorized to reproduce and distribute reprints for government purposes, notwithstanding any copyright notation thereon. The support provided by the State of North Dakota through the EPSCoR – PDC Program is also acknowledged.

Handle-Assisted Packaging (H-AP): The work in this area was supported in part by the State of North Dakota, Department of Commerce Technology Based Economic Development (TBED) Grant Program. The collaboration of our colleagues from Mühlbauer GmbH, Roding, Germany was very important for the success of this work. The author is also immensely grateful to his colleagues from Uniqarta – R. Kliger, Y. Atanasov and B. Scholz without whom this work would not be possible.

Smart Paper: The support provided by K. Doelle, SUNY Syracuse, NY and by our partners in a number of paper companies was invaluable. The team from Uniqarta –Y. Atanasov and B. Scholz – spent endless hours making sure the idea of integrating two very different technologies – the ancient papermaking technology and the innovative ultrathin flexible hybrid electronics – becomes a reality.