Review and future of active disassembly

Abstract

Disassembly has been widely accepted as a disadvantageous end-of-life activity, but with increasing pressures from directives, such as waste on electrical and electronic equipment, and with increasing pressures to become sustainable disassembly is becoming necessary. Current disassembly methods, including both manual and automated disassembly, need improvements to meet this necessity. This paper will introduce the improvements needed and suggest through literature the validity of active disassembly (AD) to provide these improvements. Past and current research will also be considered to provide a future path for AD. This future path for AD will consider a collaborative effort to solve problems with materials, environmental triggers and costs seen with AD.

Keywords:

• active disassembly
• end-of-life products
• design for disassembly
• disassembly embedded design
• active disassembly using smart materials

Introduction

Electrical devices have become a staple in homes across the world leading to an increase in the sales for such items as televisions, computers, cell phones and other similar electrical devices worldwide. What to do with this increasing number of electrical devices has become a popular topic for both industry and research alike. This increase has also provided regulations for life management and subsequent government directives for treatment of end-of-life (EoL) products. Directives, such as the waste electrical and electronic equipment (WEEE), have been hallmarks in the life management of electronics in Europe.

The WEEE directive strives to change the design of new products to promote efficient dismantling and places the EoL responsibilities solely in the hands of the original equipment manufacturers for both new and old electronics. These responsibilities include: retrieving the EoL item at no charge to the consumer, treatment of the given item, recovery and removal of hazardous materials and reuse/recycle of components and materials. The recovery/removal and reuse/recycle aspect of the WEEE directive is especially important. As of 31 December 2006, the WEEE directive has called for the limited removal and treatment of polychlorinated biphenyls, mercury containing components, printed circuit boards (PCBs) bigger than 10 cm², cathode ray tubes and liquid crystal displays (LCDs). The mode for removal of the called for components by the WEEE directive is disassembly (European Union 2003).

Besides disassembly, the other EoL options for e-waste include: landfilling and recycling by shredding. Basic disposal or landfilling of electronics has been scrutinised because of land shortages and the damage it can cause to the environment (Furuhjelm et al. 2000, Mohite and Zhang 2005). Landfilling also prevents the recovery of precious and recyclable materials (Qian and Zhang 2003). Recycling is an EoL option that will try to recover materials from the electronic waste. Recycling of electronic waste is primarily done by shredding and separating. The shredding process is very time consuming (Cui and Forssberg 2003), it limits the recovery of materials (Willems et al. 2006) and it has a variety of safety issues (Munford 2005). Along with the disadvantages of current EoL options and with constraints of directives, such as the WEEE, disassembly is becoming a much more favourable if not mandatory part of the EoL processing of electronics. In the past, disassembly has been considered as a very time consuming and expensive affair. Manual disassembly is very costly due to the inefficient disassembly design for many products, which ultimately increases the time to disassemble a product and thus the labour cost (Duflou et al. 2006). Automated disassembly requires little labour cost and research has shown that it is increasing in flexibility, but its flexibility is still limited and a large financial investment in machines must be made (Scholz-Reiter et al. 1999). Due to these inefficiencies, due to the optimal recovery of materials and due to legal constraints, improvement of disassembly methods is becoming an interesting topic for current researchers.

Of these improvements in disassembly methods, active disassembly (AD) seems to be at the forefront. AD uses innovative components within the design of the product to promote a self-disassembly of the product. While this is a novel idea, there have been problems keeping AD from being used commercially. This paper will give a brief overview of current disassembly methods and their problems, discuss how AD has evolved from design for disassembly (DfD) and disassembly embedded design (DED), consider AD in past research and the future improvements that must be made to AD to make it a viable commercial option.

Current disassembly methods

Disassembly is the recovery of materials by separating components in a reverse assembly order (Langerak 1997). Partial disassembly has been briefly discussed as a way to remove hazardous materials from electronic devices before disposal or recycling. When incorporated with shredding, partial disassembly will help increase the recovery rate of materials of a product. Full disassembly, on the other hand, will promote not only more efficient recycling of unusable components and products, but will promote the possible reuse or remanufacture of components and products (Linton 1999, Tanskanen 2003). How a product can be disassembled either fully or partially is a determinant on three distinct methods: manual disassembly, automated disassembly and AD.

Manual disassembly and automated disassembly

Manual disassembly seems to be the most popular method used in industry today (Kopacek and Kopacek 1999). It uses manual labour to take apart a product. Operators will incorporate simple to tailored made tools and machines to assist in manual disassembly (Boks et al. 1996). While manual disassembly is very flexible because with the proper tools and equipment anything can be disassembled, it is only cost efficient for a low proportion of input material. This is due to the intensive time and resulting labour cost required (Chiodo et al. 1998a, Kopacek and Kopacek 1999, Scholz-Reiter et al. 1999, Chiodo and Boks 2002, Tanskanen and Takala 2002, Duflou et al. 2006). Automated disassembly uses automation or robotics to disassemble products. The use of automated machines and robotics increases the efficiency in which products are disassembled, but the flexibility of disassembly is lost and capital cost can be high (Langerak 1997, Chiodo et al. 1998a, 1999b, Hussein 2003).

Both manual and automated disassembly methods have the same one-to-one disassembly process. This one-to-one action is described by Willems et al. (2006) as one or more actions which will disassemble one component of a product. For example, unscrewing bolts for the housing of an electric motor will only partially disassemble the electric motor. Other bolts, screws and snaps must be dealt with to fully disassemble the motor. This one-to-one disassembly process shows the inefficiency in current disassembly methods (manual and automated) and is the main reason why mass shredding or basic disposal is the norm in EoL scenarios (Duflou et al. 2006, Willems et al. 2006).

DfD and DED

To increase the efficiency of the disassembly process, research ideas and concepts have evolved from simplifying the one-to-one disassembly process to instituting specific design features to create a one-to-many disassembly process. Simplifying the one-to-one disassembly process is seen mainly through following the DfD principles. The DfD principles include:

1.	replacing difficult to remove fasteners, such as screws and glue, with easily disengaged snap fits;
2.	promoting proper design for handling and cleaning of the product, so at EoL component separation is not complicated from excess grime or dirt;
3.	standardising subassemblies and components into disassembly modules; and
4.	selecting materials that are easily recycled (Zhang et al. 1997).

Simplifying the one-to-one process by following DfD principles will increase the efficiency of disassembly, but to optimise disassembly a one-to-many disassembly process must be incorporated.

In direct contrast to a one-to-one disassembly process, a one-to-many disassembly process will remove multiple components with one disassembly action (Duflou et al. 2006). DED promotes a one-to-many process by incorporating special separation features designed specifically for a product's disassembly. These special separation features will be triggered or activated into a disassembly action by a single environmental change. Research has discussed four environmental triggers for DED including: mechanical, thermal, electro-magnetic and electrical. For each trigger, the separation feature is different. This provides many options for design with DED, but it can also make design work very intensive leading to an expensive and non-standardised design process. Intensive design work has kept DED from being widely used in practice. Another problem with DED extends beyond the design of the product to the disassembly process for the product. With the specialised separation features, the proper application of the disassembly trigger becomes essential for the product to disassemble. The complexity then becomes increased for DED because added variables, such as how and where the disassembly trigger is applied, need to be determined for the successful disassembly of the product. In theory, however, DED has become a major area of focus and has lead to the development of AD.

Active disassembly

As stated above, AD can be seen as a continuation of DfD principles and DED principles. AD separates itself from DED by relying on reactive elements that can be used in a generic fashion meaning they can be incorporated in any design without extensive planning. This absence of extensive planning is extended specifically to the application and positioning of the disassembly trigger. AD elements will react to a ‘global’ disassembly trigger meaning how and where the disassembly trigger is applied will have no effect on the self-disassembly process.

To illustrate this difference between AD and DED, consider a pnuemo-element from AD and a pneumatically activated DED snap-fit (Figures 1 and 2). Both of these devices rely on a drastic change in air pressure to activate. The major difference between the two is the DED snap-fit which will release with an increase in air pressure at a localised position. This is in direct contrast with the pnuemo-element, which will activate with a ‘globalised’ increase in pressure.

Figure 1 Example of pneumo-element.

Figure 2 Example of a DED for a pneumatically activated snap-fit: (A) during use and (B) during disassembly.

Besides pnuemo-elements, AD techniques contain a variety of other disassembling elements including: freezing elements, soluble elements, hydrogen storage alloy elements and SMM elements (Duflou et al. 2006).

Freezing elements rely on the expansion of water when frozen. The water is kept in a membrane placed inside a product and once frozen will separate the housings or components of the product (Figure 3). Since the concept of the freezing element is based on the expansion of water, special consideration needs to be made with the dimensions of the element and conditions during disassembly because both of these issues will affect how much the element expands. Making these considerations usually produces relatively large freezing elements, which are unsuitable for smaller products (Duflou et al. 2006).

Figure 3 Example of freezing element.

Pneumo-elements are releasable fasteners, which are activated by a pressure differential. The pneumo-element is an enclosed body filled with compressed air. The enclosure of the pneumo-element is flexible, so when the pneumo-element is placed in a high pressure environment it will shrink and release (Figure 1). Pressure is a very controllable trigger making the occurrence of accidental triggering a relatively low possibility. Special safety factors do need to be developed with pneumo-elements to account for the varying release forces especially for old connections. Considerations for the ambient pressure must be made especially for non-pressurised environments. Like the freezing elements, the pneumo-elements can occupy a large footprint within a product (Duflou et al. 2006).

Soluble elements will hold a fastener when in a dry environment. The disassembly action is seen when in the presence of a fluid; the soluble element will dissipate releasing the fastener (Figure 4). Depending on the soluble material, temperature of the fluid to dissolve the element can be either high or low. This provides a safety factor known as a double trigger because two environmental changes or triggers must occur for the element to dissipate, the presence of water and high temperature or vice versa. Double or multiple triggers are a definite need for AD, but the double trigger for soluble elements has been overlooked because degradation to the element can occur very easily in humid environments (Duflou et al. 2006).

Figure 4 Example of soluble element.

Hydrogen storage alloy elements employ a chemical reaction to separate bonded components. The components are bonded together using a hydrogen storage alloy. Once the hydrogen storage alloy is placed in a high hydrogen environment, the alloy will break and release the bond (Suga and Hosoda 2000). Since this AD element is only used for bonded components, its applicability for separating product housings, assemblies and sub-assemblies is questionable.

AD with SMM elements or AD using smart materials (ADSM) is by far the most researched area within the AD realm. ADSM incorporates the use of SMMs within the design of a product to promote a disassembly action much like DED. Past DED experiments have incorporated special disassembly features made of shape memory alloys (SMAs) to disassemble a product. Once the SMA disassembly feature is heated, a memory effect will occur and the alloy will distort separating components of the product (Chiodo et al. 1999a, 2002). While this is an example of a smart material being used within the DED realm, AD separates itself from DED by being more flexible and less design intensive by using generic SMM fasteners and housings which can be used in many current and future products (Duflou et al. 2006). As with the special separation features in DED, the fastener or housing will stay dormant during the use phase of the product and once activated will cause a self-disassembly process (Tanskanen 2003).

SMMs change their mechanical properties in relation to a drastic temperature change at a certain transition temperature, T _x . At T _x , the SMM will perform a shape memory effect (SME) or exhibit a drastic change in shape (Jones et al. 2004). SMM are separated by SMAs and shape memory polymers (SMPs). Each of these materials performs their SME by differing means. SMAs will deform owing to a stress-induced martensitic transformation. Essentially, the SMA will be deformed above its transition temperature and while cooled held in this deformed state. When cooling is complete, the SMA will remain in its deformed state until it is heated above its T _x . Once heated above T _x , the SMA will return to its natural shape. The stress-induced SME from the SMA can produce high levels of force to separate or pop open components making SMAs suitable as a separating actuator.

An SMP will go through a similar shaping process, but the SME is due to a special cross linking of the molecular chain (Otsuka and Wayman 1998). The micro-Brownian movement of the molecular chain essentially is a random fluid-like movement allowing easy deformation of the SMP, but if the temperature of the SMP is lowered below the T _x , the micro-Brownian movement will cease. Once the micro-Brownian movement has stopped, the SMP will be rigid, which means the oriented molecular chain and deformation will also be fixed (Tobushi et al. 1993, Hayashi et al. 1994). The SME is most often characterised by a drastic loss of rigidity with SMPs; however, there are cases where a major change in shape can occur, but the force of the SME is not significant. This makes SMPs useful as a releasing fastener and/or housing.

AD past research

Past research for AD has primarily focused on retrofitting existing electronic devices with generic SMM fasteners to promote a disassembly sequence. This research has been successful for the most part, but a number of issues have arisen. These issues deal mainly with the mechanical properties, material properties and cost of the SMMs used. The research performed by the Katholieke Universiteit Leuven, the Research Center for Advanced Science and Technology (RCAST) group at the University of Tokyo, Cleaner Electronics Research group at Brunel University and the Nokia Research Center (NRC) have done well to identify these issues and determine possible solutions to make AD commercially accepted. These institutions have also begun efforts to design for AD (DfAD) by studying the feasibility of AD and creating some DfAD guidelines.

Katholieke Universiteit Leuven

Katholieke Universiteit Leuven has been the second group that has extensively studied the feasibility of AD. The feasibility studies from the Katholieke Universiteit Leuven group have focused on one-to-many disassembly processes seen in both DED and AD. The group supports AD over DED because DEDs are very application and product specific, are not very manufacturable and call for intensive design work. AD designs, on the other hand, are much more flexible, are easily manufactured and can be based on previous product designs. The main disadvantages of AD designs, however, are due to the triggering mechanisms in AD. For the most part, AD elements only use a single environmental change as a trigger making accidental triggering a real possibility. Extra costs must also be made to design the AD elements. Since these costs are not explicitly seen, Willems et al. (2006) raises doubt in the actual benefit gained from using AD.

An evaluation of each disassembly technique is also made by Duflou et al. (2006) for both DED applications and AD applications. Each application was graded on five criteria: time, fixture, localise-visible-access, assembly effort and level of integration. Time correlates to the disassembly time. Fixture criterion classifies the type and number of fixtures used to disassemble if any. Localise-visible-access relates to ease at which to access disassembly points for DEDs. Assembly effort scores the addition of any parts needed for disassembly. Level of integration concerns the depth at which AD element will disassemble the product. A composite score was gathered from these five criteria, and AD with SMPs was shown to be the best among the group (Duflou et al. 2006).

University of Tokyo

The RCAST at the University of Tokyo has performed studies and experiments concerning hydrogen storage alloy elements and other debonding elements. With a concentration on hydrogen storage alloy elements and other debonding elements, the RCAST group introduced the concept of reversible interconnections. Reversible interconnections is a concept stating that anything bonded together should be able to be separated with ease and should not call for the destruction of the bonded joint. The group gives examples of where and how reversible interconnections should be applied. These examples include debonding of Al–Si, Al–LaNiAl, Al–Fe and Cu–Al₂0₃. The group suggests that debonding of Al and Si can occur with interface reactions with hydrogen radical or hydroxyl group. Al–LaNiAl can be debonded as in the case of any hydrogen storage element because LaNiAl will be pulverised in a high hydrogen environment. During bonding of Al and Fe, a brittle intermetallic compound is formed to hold the two metals together. Once this bond is heated to a very high temperature, this brittle bond will fail without applied external force. The same principle is used with Cu–Al₂O₃ bonds except separation will be made by the great mismatch of the thermal expansion coefficients of each material (Suga 1999, Suga and Hosoda 2000).

These examples would be applicable to separate components from PCB and integrated circuit substrates. Separation of the silicone substrate is another point of interest for the RCAST group. To do this, the RCAST proposes placing water between wafers before bonding. The water is then evaporated to disassemble the Si wafer. The group models the parameters needed to successfully disassemble the wafer-based critical pressure needed to separate the wafers (Kasa and Suga 1999).

Brunel University

Brunel University has been the leader for AD research or more specifically ADSM research. The Cleaner Electronic Research group at Brunel has done extensive studies with AD with both SMAs and SMPs. With these studies, the group has also performed many feasibility studies concerning ADSM.

The SMA experiments performed by the Cleaner Electronics Research group began in the late 1990s and early 2000s. The results from their experiment showed the flexibility and effectiveness of ADSM for electronics. The first study performed by the group incorporated SMA actuators to separate ICs from PCBs and separate the housings of the PCBs. Separation of the ICs from their PCBs proved unsuccessful because the output force of the SMA actuators was insufficient. Separation of the product housing, however, was successful in all cases studied but one (Chiodo et al. 1997). In Chiodo et al. (1998b, 2002), the Cleaner Electronics Research group uses a variety of SMA devices including NiTi discs, rods and coils and CuZnAl coils to disassemble many electronic devices. These SMA devices were fabricated and strategically placed inside existing electronic devices. The electronic devices were then heated to actuate the SME in the SMA coil causing a self-disassembly process. Once heated, the SMA coil, rod or disc expanded. This expansion forced the housing or components within the electronic device apart. A camera, calculator, electronic game controller and PC mouse were just a few of the items that were successfully disassembled. While Chiodo et al. (1998b, 2002) were both successful in the disassembly of a majority of the electronic devices tested, there were a few devices that were not successfully disassembled stating that with some minor changes to the product design all the devices tested could be disassembled. If these changes were not made the authors believed more destruction could come to the product eliminating some of the benefits of the ADSM process. Another issue of concern to the Cleaner Research group was the issue of accidental triggering due to a high and sustained ambient temperature.

Chiodo et al. (1998b, 2002) both used a heat to initiate the SME by either placing the product in a hot air chamber or hot water bath. In Jones et al. (2004), a different triggering approach is taken with SMA. Jones et al. (2004) uses the residual power left in EoL cell phone batteries to induct heat for triggering a SME in a SMA ‘muscle wire’. In the DED experiment, the SME was used for cell phone battery ejection. The muscle wires were setup in two specific ways for battery release. One design had the muscle wire hold back a latch that released a biasing spring. The muscle wire would release the latch, which released the biasing spring, ejecting the battery from the cell phone. In the second solution presented by Jones et al. (2004), the muscle wire is used to pull back a moulded bell crank. The bell crank will tilt the battery up ejecting the battery from the cell phone. EoL cell phone batteries were also tested, and it was determined that the residual power left in these batteries would be enough to activate the SMA muscle wire. Along with studying residual power from cell phones for activating a SME, Jones et al. (2004) also explore the residual power in EoL car batteries for disassembly. The results show enough power from these batteries to activate shape memory devices (SMDs) after vehicle depollution. Once successfully depolluted, the SMD will be activated by the residual power in the battery, and parts can be easily separated along with the battery for subsequent recycling processes.

Other ADSM studies performed by the Cleaner Research group at Brunel University focused on the use of SMPs. The first noted studies used a SMP releasable fastener in the form of holding brackets and of a compression sleeve. The releasable fasteners were placed in telephone assemblies to disassemble large product housings. The holding brackets replaced conventional screw assemblies and would release their hold on the housing once they were above the transition temperature. The group designed SMP mechanical property loss (MPL) screws to disassemble Nokia 6110 and Nokia Populus cell phones (Chiodo et al. 1999a), a game controller, a clock radio and a CD player (Chiodo et al. 1999b). MPL screws will become very pliable when the SMP is above its T _x temperature and will essentially lose their threading (Figure 5). These MPL screws replaced existing fastening elements in these electronic devices, and with the aid of biasing springs, the electronic devices would open up and disassemble when heated above the T _x temperature. As with the group's experiments with SMA actuators, some minor alterations had to be made to the product design to account for the biasing springs (Chiodo and Boks 1999, Chiodo et al. 1999b).

Figure 5 Phase relationship of SMP MPL screws: (A) below T _x , (B) slightly below T _x and (C) above T _x .

The SMP MPL screws and SMA actuators were used by the Brunel group to retrofit existing products. Experiments with retrofitted products have not been the sole effort of the group. The Brunel group has designed SMP brackets used to separate a LCD from its housing. The SME with the SMP bracket is different from SME with the MPL screws. Instead of losing all rigidity the bracket will contort and essentially separate itself from the LCD display (Chiodo et al. 2000).

Feasibility studies concerning AD have also been a point of interest to the Brunel group. These feasibility studies have been helpful in developing design guidelines for AD and developing future work for the advancement of AD. For design guidelines, the designer has many options to choose from based on the disassembly action of the actuators, material of the actuators and triggering method for the actuators. The disassembly action of the device is determined by the SMM used. SMAs can be used in situations requiring a force for separation, while SMPs can be used as releasable fasteners. The SMM will also determine the material properties of the actuator. SMAs and SMPs have varying transition temperatures meaning sequencing the disassembly of the device can be based on the heat distributed to it. Along with the transition temperature of the device, the mechanical properties of the device will be determined by the SMM used. For the most part, SMMs have low mechanical strength, when compared to their regular engineering counterparts, making the SMMs only suitable in non-supportive roles in the product's design. Heat is the main triggering method for the SMM, and how the object is heated is vital because improper heating can destroy the components and parts of the product being disassembled. Heating choices include: convection, immersion, microwave, induction and infrared. Each heating method has its own advantages and disadvantages that must be weighed by the designer to ensure the products remain recoverable after disassembly (Chiodo et al. 1998a, Chiodo and Boks 1999, Warburg et al. 2001, Chiodo and Boks 2002).

Nokia Research Center

The NRC has provided equipment and knowledge to the Cleaner Electronics Research group at Brunel University and has also performed its own case study and feasibility study concerning ADSM. The centre's case study focused on the use of a SMA coil in a disassembly system for a prototype cell phone cover (Figure 6). The NRC provided details in the design process for their mechanisms by elaborating on the specific design decisions. The design process was kept standard with the exception of judgement criteria for the disassembly system. The criteria tested include: the functionality and reliability throughout the life of the system, reaction temperature and speed for the system, size of the system, simplicity of the system, cost for the system and service ability for the system. With these criteria, design features for the system and mechanism were clearly defined and justified by the NRC.

Figure 6 Disassembly system for a cell phone cover using a SMA coil. Left, engaged snap-fit on SMA coil. Middle, SME after heating coil above transition temperature. Right, removed snap-fit.

The mechanism of the system is very simple. A trained SMA coil is placed in a metal housing with a heating stud. Once the heating stud is heated the SME in the coil will occur, which in this case will be a constriction of the coil releasing the snap. The heating stud is the target for a laser heating apparatus. Laser heating on the stud was chosen to ensure a fast reaction time and to reduce energy usage from heating the entire product. The designed system reacted well to testing. Disassembly occurred in only 2 s (98 s faster than with manual disassembly). While the NRC was concerned about the cost and mechanical properties of the material, the overall conclusion drawn was that ADSM could be a feasible technology once it reaches maturity (Tanskanen and Takala 2002).

The NRC has also done conceptual work with ADSM. The centre has designed a complete cell phone prototype (Nokia 5510) that can be fully disassembled with a multitude of ADSM mechanisms. The Nokia 5510 prototype was designed to be disassembled into four sections including: the PCB, LCD, covers and battery. These four sections can be disassembled in order because the SMMs can actuate at different temperatures (Tanskanen 2003).

Future of AD

Some problems with AD have already been introduced especially for non-ADSM techniques. These problems have kept these AD elements (freezing, pneumo, soluble and hydrogen storage alloy) from being universally accepted and commercially viable. These deficiencies have also made ADSM the most promising option for future AD research. For this reason, future studies will focus on improvements of materials and design to solve problems with ADSM. These problems include:

1.	Mechanical properties. Both SMA and SMP have very low mechanical strength when compared to regular engineering materials (Hussein and Harrison 2004, Ji et al. 2006).
2.	Triggering mechanisms. Past research has primarily used heat as a trigger for the SME in the SMM. In many cases, temperature is a very uncontrollable medium, which means accidental triggering is likely to occur.
3.	Cost issues. The prohibitive cost of SMMs along with the lack of explicit costs for training and recovering the SMMs have been a contributor to their lack of use in commercial applications (Chiodo et al. 2001, Hussein and Harrison 2004).

How these problems are solved is of concern for the commercialisation and industrial use of AD. The options that are currently being developed in literature include the use of design methodologies, such as Teoriya Resheniya Izobretatelskikh Zadatch (TRIZ), and design analysis tools, such as finite element analysis (FEA) to analyse and design of experiments, improve and innovate AD elements.

Design methodologies: innovations, improvements and challenges

Idea generation and problem solving

Past studies with AD have focused on the feasibility of AD elements and have identified problems with these technologies. What is drawing more interest with AD research has been the institution of design methodologies like TRIZ and other brainstorming techniques to solve some of the problems and provide further innovative ideas for AD elements (Chen and Chen 2007).

TRIZ is the Russian theory for inventive problem solving and is a systematic way to incorporate innovation into the design process. TRIZ is based on an extensive patent study performed by Generich Atshuller. Through Atshuller's study, a set of innovative patents were identified, and from these patents, commonalities of design problems and their solutions were defined. The design problems from this study were formed into a contradiction (Shirwaiker and Okudan 2008). For example, the basic contradiction for using AD instead of DED can be defined as instituting an easily disassembled element within the design of a product without increasing its complexity. This contradiction would correspond to an improving parameter, repairability/disassemblability and a worsening factor, complexity. With these parameters clearly defined, the TRIZ contradiction matrix would then be consulted. The TRIZ contradiction matrix took the commonalities of design solutions from Atshuller's study and organised them per improving and worsening parameters (Shirwaiker and Okudan 2008). For this example, the solution, which corresponds to the principle of AD, from the TRIZ contradiction matrix is ‘changing the aggregate state of the object’ meaning the state of the element would have a state for assembly/use and a state for disassembly. This solution is exactly what AD relies upon.

Besides this generic explanation of why AD should be used over DED, possible solutions for design problems with AD elements are being presented with TRIZ. Chen and Chen (2007) present a few examples of basic problems from AD studies and by using the TRIZ contradiction matrix provide possible solutions. Beyond solving basic problems with AD, a means of idea generation can be made with another TRIZ tool, a substance-field or Su-field analysis (Chen and Chen 2007).

Su-field analysis is a tool used in TRIZ to model the substances and fields of a system. A substance would be any physical object within the system, and a field would be the energy, which corresponds to those substances, to perform the given functions of the system (Shirwaiker and Okudan 2008). For AD, there are two substances, which are the components or parts being joined, the field acting on these substances is responsible for causing the components to separate or disassemble (Figure 7). The most common field used in AD is a thermal field for activating SMMs, but this does not have to be the case and this is where ideas are being generated.

Figure 7 Generalised Su-field model for AD.

Duflou et al. (2006) illustrate this idea generation for AD by presenting existing and possible fields for AD elements. In this overview, Duflou et al. (2006) include the following disassembly fields: mechanical, electrical, chemical, thermal, magnetic, light irradiation and biological. The mechanical field AD elements would include previous pressure-based elements like pnuemo-elements, but would also extend to new elements where other mechanical forces, such as centrifugal forces, acceleration/deceleration forces and vibratory forces, are used to case disassembly (Duflou et al. 2006).

For the electrical field AD elements, Duflou et al. (2006) introduce new materials for AD. These materials are electroactive polymers (EAPs). EAPs are like SMMs in that once in the presence of an electric stimulus will change shape. The mechanical properties of EAPs are like SMPs because they too have a relatively weak strength, and EAPs are like SMPs in that they do not exhibit high stresses during their shape change and are more warranted as releasable fasteners rather than actuators (Leo 2007). While there have yet to be any practical examples EAPs being used in the AD realm, EAPs have been used in DEDs. One such example is a releasable fastening system for General Motors. Seen in Figure 8, a knob element is composed of EAP in the form of an ionic polymer-metal composite. This knob is a part of an electrical circuit and will change shape for assembly or disassembly based on the polarity of the power source (Momoda et al. 2005).

Figure 8 General Motors' releasable fastening system using EAPs: (A) connected circuit shapes fastener for insertion, (B) circuit broken for use stage and (C) polarity reversed for removal of fastener.

Another material that exhibits shape change in the presence of an electrical stimulus is SMP nanocomposites. SMP nanocomposites are basic SMPs that have incorporated carbon nanotubes (CNT) within their polymer matrix. With the addition of the CNT, these SMP nanocomposites show the ability to induce a SME with a change in electric field. These SMP nanocomposites also exhibit better mechanical properties, such as strength and modulus. Another benefit, which may be possible with SMP nanocomposites, is the ability for this material to exhibit a double trigger from both electrical and thermal fields. This is something that is very needed with AD to prevent cases of accidental disassembly. Various research results for SMP nanocomposites are summarised in Table 1.

Table 1 Literature review of SMP nanocomposites.

Article	Summary	Result
Gall et al. (2002)	Created SMP nanocomposites using a commercial SMP resin and nanometer scale SiC particles	Micro-hardness and elastic modulus increased up to a factor of 3
Cho et al. (2005)	Synthesised shape memory polyurethane composites with CNT	Electro-active SME possible. Increased modulus and tensile strength
Paik et al. (2006)	Developed conducting shape memory polyurethane actuators by adding CNT to shape memory polyurethane	Improved dispersion of CNT allowing for further improvements in mechanical and electrical properties
Sahoo et al. (2007)	Tested effects of adding and dispersing CNT to shape memory polyurethane block copolymer	Significant improvement in mechanical properties. Modulus increased by 200% and tensile strength increased by 37%
Koerner et al. (2005)	Tested effects of adding and dispersing small amounts of CNT to thermoplastic elastomer morthane	Increase in conductivity, yield stress, stress at break and modulus. No loss in ability to stretch beyond 1000%
Goh et al. (2003)	Created phenoxy nanocomposite by melt mixing phenoxy with CNT	Increase in modulus by more than 600 MPa. Reduction in T x

For the chemical field, existing elements, such as water soluble elements and hydrogen storage alloys, would definitely apply, but Duflou et al. (2006) suggest the use of other chemically reactive and bioactive materials to be investigated for AD. The applicability in AD for these materials would be the property change of these materials when immersed in a different environment. Materials in the chemical and biological field have shown special properties when placed in different environments whether it be due to the pH level, hydrogen levels, moisture level, etc. and what is important for AD is to harness these special properties for use in self disassembly.

Magnetic field elements have been used in simple applications for fastening and security systems. These elements can be classified as magnetic field DEDs because of the dedicated and structured form in which the magnetic field must be applied, but advances in magnetoactive materials may provide avenues where the magnetic field can be applied in the AD realm. To illustrate some of the past DEDs using a magnetic field, consider the magnetically releasable target lock (Figure 9) and the magnetic loop and hook fastening system (Figure 10). The magnetically releasable target lock is commonly used as a security device at many retail stores. How the device operates is quite simple. A pin will be inserted into a locking mechanism that consists of a set of flanges. The flanges will not relinquish the pin upon a pulling force. For release, a magnetic field must be applied to the flanges causing them to turn and release the pin (Minasy and Olszewski 1991). How the magnetic field is applied to the device is very important for successful release and is the major reason why this device can be considered a DED device. The magnetic loop and hook fastening system is an interesting system in that it has spawned another field for use, a thermal field, and with this thermal field has found a way into the AD realm. The basic premise of the magnetically triggered system is upon a directed magnetic field; the hooks will straighten and be easily removed (Ulieny and Golden 2004). Again as with the magnetically releasable target lock, this system can be seen as a part of the DED realm because the direction and application of the magnetic field and the orientation of the hoops and loops within the product will determine the success of disassembly. By replacing the magnetic field with a thermal field and replacing the hoops with a SMM, this hoop and loop system becomes an AD system because it is no longer dependent on the direction for which the field is applied or the orientation of the hoops and loops (Powell and Browne 2006, Vokoun et al. 2009).

Figure 9 Magnetically releasable target lock: (A) before insertion, (B) insertion and use, (C) magnetic field application and (D) removal.

Figure 10 Magnetic loop and hook fastening system: (A) view of single magnetic hook, (B) before assembly, (C) assembly and use and (D) magnetic field application and disassembly.

While current designs for magnetically triggered disassembly are in the DED realm, newer materials may provide a means for the magnetic field to be applied in the AD realm. Magnetoactive and magnetostrictive materials, like their electroactive counterparts, will exhibit special properties upon application of a magnetic field. These properties vary from material to material. Magnetic SMAs, for example, will act much the same as regular SMA, but instead of a thermal trigger, a magnetic field will cause the SME and a significant force from actuation (Ullakko 1996, Tellinen et al. 2002). Other materials, such as magnetoactive elastomeric composites, incorporate magnetically active materials within a polymer matrix to give the polymer special magnetic properties (Sasikumar et al. 2006). Another set of magnetically triggered materials will change states. Magneto-rheological materials will change from a viscous liquid to a solid-like material when in the presence of a magnetic field (Jolly et al. 1999). Like their smart material counterparts, magnetoactive materials do have some drawbacks, which will limit their applications, but with innovative ideas these materials may solve some of the problems exhibited in AD research.

The last group of materials of interest for AD are triggered by light irradiation. Light has long been an energy source which research has tried to harness. This is evident by the amount of work concerning solar panels and photovoltaics in which light energy is converted into electrical energy. A new group of materials known as photoresponsive polymers are currently being studied that turn light energy into mechanical energy or movement. Examples of these photoresponsive polymers include infrared active shape memory nanocomposites (Koerner et al. 2008) and photo-induced phase transition polymers (Ikehara et al. 2002). For the future integration into AD, special consideration in design should be made to ensure the triggering mechanism is simple and reliable. This would be especially true in cases where the element is within the product assembly and is otherwise covered from the light source.

While triggers and materials may innovate the way elements are designed for AD, there will be some challenges. This seems to be especially true for mechanically, electrically, magnetically and light irradiation triggered elements where extra effort in the design of the element, the design of the product and design of the field application process is required to ensure disassembly. It can be argued that the current designs that exist with these triggers, especially magnetically and electrically triggered, are not designed for AD, but are instead DEDs. To solve this issue, more research in materials and field application must be performed to determine if the design of the element and product will allow for a generic application of the field. While there is an added flexibility in design because many different materials and triggers maybe used, the true root of the triggering issue is not addressed with these materials, excluding maybe SMP nanocomposites. What must occur for the elimination of accidental triggering is the concept of multiple triggering. SMP nanocomposites may provide a means for this multiple trigger, but other solutions may exist, which provide a means of a multiple trigger, but also reduce the likelihood of accidental triggering. Solutions such as using multiple smart materials within the design of the disassembly element to promote double or multiple triggers would fall in this category. Other solutions, such as using design analysis tools and methods to make the designs more robust and less sensitive to environmental conditions, would be another avenue for designers and will be discussed in the following sections.

Other design tools and studies

For other design methods, consider the work performed by Hussein and Harrison (2004, 2008). In both of these articles, standard engineering polymers, which promote a SME, are used in the design of an AD snap-fit. Design tools, such as designed experiments and classical beam theory, are used in the design and analysis of the snap-fits (Hussein and Harrison 2004, 2008). The concept of using this SME to perform AD is currently an interesting option because it looks to solve both mechanical and economical problems with ADSM. Standard engineering materials are already used in many applications meaning their mechanical strength and costs are already optimised. The standard engineering materials will also have a higher transition temperature than their shape memory counterparts making them less likely to trigger from ambient temperatures. Looking at SMPs, in particular, their transition can occur from 25 to 60°C while for a basic thermoplastic, such as an ABS/polycarbonate blend, the range is from 120 to 140°C.

Existing design equations, for fasteners and sophisticated ways to model these fasteners, provide another avenue for solving mechanical problems with current AD elements. The work performed by Shalaby and Saitou (2008a, 2008b) is an illustration of the use of such equations and models. For their work, it is shown by slightly changing the design of AD elements provisions for more control in disassembly can be made helping to solve issues with accidental triggering. Their concept deals with the application of Screw Theory to increase the reliability of SMP locator snaps and decrease the likelihood of accidental triggering (Shalaby and Saitou 2008a, 2008b). Disassembly will only occur when the temperature is sustained at the transition temperature (Figure 11).

Figure 11 ADSM releasable snap design with more control: (A) T≪T _x , (B) T < T _x , (C) T = T _x and (D) T>T _x .

This concept is quite simple, but the modelling of the system can be quite complicated because the constraints of the material must be taken into consideration to ensure the desired behaviour is reached. Luckily tools, such as FEA, have already been incorporated to show the validity of smarter designs with more control (Shalaby and Saitou 2008a, 2008b, 2009). Currently, these smart designs with more control are only focused on eliminating accidental triggering and increasing disassembly reliability, but they do not attack other deficiencies with ADSM most notably mechanical limitations with SMMs.

FEA is a powerful design tool and has been used in other AD assessments including Willems et al. (2007). In Willems et al. (2007), topology optimisation with FEA are used to model the assembly, use and disassembly of the pressure element. By modelling in each of these stages, the design of the pressure element can be optimised to meet the requirements placed on it throughout the entire life cycle. More importantly, the simulation of this element will provide the necessary decision criteria for implementation (Willems et al. 2007).

Conclusion

The basis of AD is to provide a means of disassembly that is both flexible and efficient. AD has shown its validity in the research realm as an effective way to disassemble products. However, material issues, triggering issues and cost issues have kept AD from being commercially accepted. Solutions for these issues have been presented, but alone these solutions only target a fraction of the issues. What is then needed is a combination of these solutions. A combination will call for a collaborative effort across disciplines including mechanical, chemical and industrial. Chemical input is needed to make AD materials stronger and more controllable with multiple triggers. Mechanical input is needed to make AD designs more robust and reliable while in its use phase and while in its disassembly phase. Industrial input is needed to bridge the gap between materials and design to make the manufacturing and use of AD more affordable. Only with this collaboration is it likely that AD will be widely accepted for commercial use.

Acknowledgements

The authors would like to gratefully acknowledge financial support from the Advance Manufacturing Laboratory at Texas Tech University and the Texas Tech Industrial Engineering Department. This paper is based on research funded by the Texas Tech Summer Dissertation Award.