- Email: [email protected]
On the selection of shape memory alloys for actuators
Materials and Design 23 Ž2002. 11᎐19
On the selection of shape memory alloys for actuators W. HuangU School of Mechanical and Production Engineering, Nanyang Technological Uni¨ ersity, Nanyang A¨ enue, Singapore 639798, Singapore Received 21 June 2000; accepted 2 April 2001
Abstract This paper presents a series of charts for the selection of shape memory alloys ŽSMAs. for actuators. It is based on performance indices with special reference to the unique features of SMA actuators. The three most popular polycrystalline SMAs are candidates of current study, namely, NiTi, CuZnAl and CuAlNi. More SMAs may be added in to give a complete picture. 䊚 2002 Elsevier Science Ltd. All rights reserved. Keywords: Shape memory alloy; Actuator; Selection; Performance index
1. Introduction The unique shape memory behavior in some smart materials attracts a lot of attention from various communities. Among them, shape memory alloys ŽSMAs. show great potential in many applications, in particular, as novel actuators. Many alloys displaying shape memory have been found and considerable effort is still being made to discover new materials. However, the property of these SMAs varies significantly. It turns out to be necessary to have a systematic scheme for the selection of SMAs. There are some preliminary works on this topic w1,2x, which cover some concerns in real applications. The aim of this paper is to present a series of charts for the selection of SMAs for actuators, in particular, at the pre-design stage. Current discussion focuses on the three most popular polycrystalline SMAs, namely, NiTi, CuZnAl and CuAlNi. The discussion is purely based on the material properties reported in the literature. 2. Properties of SMAs A SMA is able to memorize and recover its original U
Tel.: q65-7904859; fax: q65-7911859. E-mail address: [email protected] ŽW. Huang..
shape after being significantly deformed from heating over the phase transformation temperature. This phenomenon is known as shape memory effect, Fig. 1a. On the other hand, at a high temperature, the large deformation can be fully recovered by simply releasing the applied load. This is superelasticity, Fig. 1b. What is involved in both kinds of phenomena is either transformation between low temperature phase Žmartensite . and high temperature phase Žaustenite . or reorientation among different martensite variants. It is well known that SMA is very sensitive to the exact composition, grain size, processing Žincluding heat treatment . and loading conditions w3x. The inclusion of additional elements with minute quantity may change the properties of a SMA dramatically. For instance, the addition of a small amount of copper in NiTi SMA makes the transformation temperature less sensitive to composition and hysteresis gap narrower w4x, adding a small amount of palladium into NiTi SMA, high transformation temperature Žover 200⬚C. is achievable w5x. The sensitivity of SMA seems to be an intrinsic disadvantage. On the other hand, it provides tremendous flexibility to get exactly what is wanted within a far broader range on comparison with traditional materials. Unlike ordinary one-way SMA, which can only remember the high temperature austenite shape, twoway SMA, which can remember both high and low
0261-3069r02r$ - see front matter 䊚 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 2 6 1 - 3 0 6 9 Ž 0 1 . 0 0 0 3 9 - 5
W. Huang r Materials and Design 23 (2002) 11᎐19
12
Fig. 1. Typical behavior of SMA. Ža. Shape memory effect and Žb. superelasticity.
temperature shapes, is obtainable by special training Že.g. w5x.. However, the phase transformation strain is significantly less and their stability and repeatability are uncertain. The R phase is an intermediate between austenite and martensite in some SMAs w6x. Upon cooling NiTi SMA from a high temperature, the material may transform from austenite to R phase first and then to martensite upon further cooling. The hysteresis associated with R phase is normally narrow Ža few degree or less. and the corresponding phase transformation strain upon stressing is small Žapprox. 0.2%.. Therefore, it cannot be used as actuator other than switch.
To date, the work on various SMAs is extensive. However, for most SMAs, the study is far from complete. In view of this, our discussion in this paper focuses on the three most popular SMAs, namely, NiTi, CuZnAl and CuAlNi. Only one-way SMA is considered. Monocrystalline SMAs may have a better thermo-mechanical property than polycrystalline material. Due to the difficulty in fabrication and the very high price, they are not presently of much commercial importance. Hence, only polycrystalline SMA is the interest of the current study. Material properties of each SMA are listed in Table 1. They are mainly taken from w3x and w15x.
Table 1 Main properties of typical SMAsa Žmainly taken from w3x and w15x. NiTi
CuZnAl
CuAlNi
Specific heat ŽJrKg⬚C. Thermal conductivity Ž20⬚C. ŽWrmK. Density ŽKgrm3 . Latent heat ŽJrKg. Electrical resistivity Ž106 ⍀ m. Thermal expansion coefficient Ž10y6 rK.
450᎐620 8.6᎐18 6400᎐6500 19 000᎐32 000 0.5᎐1.1 6.6᎐11
390᎐400 84᎐120 7540᎐8000 7000᎐9000 0.07᎐0.12 17
373᎐574 30᎐75 7100᎐7200 7000᎐9000 0.1᎐0.14 17
Maximum recovery stress ŽMpa. Normal working stress ŽMpa. Fatigue strength ŽNs 106 . ŽMpa. Maximum transformation strain Ž%.
500᎐900 100᎐130 350 6᎐8 6᎐8 2᎐4 Ž3. 0.5 ) 105 28᎐83 y200᎐200
400᎐700 40 270 4᎐6 4
300᎐600 70 350 5᎐6 4
) 104 70᎐100 y200᎐150
) 5 = 103 80᎐100 y200᎐200
2᎐50 400
5᎐20 150
20᎐40 300
Damping capacity ŽSDC%. Grain size Ž m.
15᎐20 1᎐100
30᎐85 50᎐150
10᎐20 25᎐100
Melting, casting and composition control Forming Žrolling, extrusion. Cold-working Machinability Cost ratio
Difficult Difficult Fair Difficult 10᎐100
Fair Easy Restricted Very good 1᎐10
Fair Difficult Very difficult Good 1.5᎐20
Normal number of thermal cycles Young’s Modulus ŽGpa. Shape memory transformation temperature Ž⬚C. Hysteresis Ž⬚C. Maximum overheating temperature Ž⬚C.
a
Additional alloying is not considered.
Ns1 N - 102 N - 105 N - 107
W. Huang r Materials and Design 23 (2002) 11᎐19
13
3. SMA actuators There are three basic types of SMA actuators using one-way SMAs ŽFig. 2.. 䢇
䢇
䢇
Fig. 2a shows a one-way actuator. The SMA element is elongated initially at low temperature and then is heated up to pull element P to the left; Fig. 2b shows a biased actuator, which is capable of moving the element P back and forth by heatingrcooling SMA element; and Fig. 2c shows a two-way actuator, which includes two one-way SMA elements. P may move back and forth by heating and cooling the two SMA elements alternatively.
Basically, SMAs may be heated up by three methods, i.e.: 䢇
䢇
䢇
By passing an electrical current through them. This method is only applicable to small diameter SMA wire or spring; by passing an electrical current through a high resistance wire or tape wrapped around SMA. This method is applicable for bulk SMA; and by hot airrwater or exposing to thermal radiation.
The bandwidth of SMA actuators is usually limited. Cooling speed is the dominating factor. Since small hysteresis gap requires small temperature variation, SMA with small hysteresis is preferred by high bandwidth application. Heat conduction is important in bulk SMA. Experiment shows that 6 ms per cycle can be realized in a switch triggered by 0.1-mm-diameter NiTi wire under free convection condition in air w7x. For larger sized SMAs, additional cooling method is required. Flow liquid can improve the bandwidth dramatically w8x. A fan can provide compressive convection for fast cooling w9x. Other technologies that have been exploited include semiconductors w10x, mobile heat sink w11x and pairs of agonistic᎐antagonistic wires w12x.
Fig. 2. Basic types of SMA actuators using one-way SMAs. Ža. One-way actuator, Žb. biased actuator and Žc. two-way actuator.
Fig. 3. A simple linear SMA actuator. Ža. High temperature and Žb. low temperature.
4. Material selection Actuators offer various performances and operate in many different ways. In terms of materials selection, there are always some special goals to meet. They can normally be presented by means of a performance index ŽPI., while the properties of materials can be presented by a variety of charts w13x. It should be pointed out that not every goal could be represented by means of a PI, in particular for those related to nonthermo-mechanical properties. For instance, in biomedical applications, the only candidate is NiTi due to its excellent biocompatibility w14x, which is lacking of in other SMAs. In practice, one should select materials from the charts according to the particular PI required by a special application. However, since these charts are not available yet, we will have to identify the most important PIs first, produce the corresponding charts and then make the choice.
Fig. 4. Transformation temperature vs. hysteresis.
14
W. Huang r Materials and Design 23 (2002) 11᎐19
Fig. 5. Density vs. output volumetric work. Ža. N s 1 and Žb. N s 100.
Fig. 3 shows the principle of a typical linear actuator. This is the simplest actuator, a weight P hangs at the bottom of a SMA wire. At a low temperature, P moves down Fig. 3a. Upon heating, SMA wire contracts and pulls P up, Fig. 3b. External energy is required only in the heating process. Despite the simplicity, in terms of performance, this actuator represents most of the concerns of SMA actuators. A constant compressive force or a constant torque may replace weight P in a real application. One has to realize the effect of asymmetry in tension and compression in many SMAs. The detail on how to deal with it is presented later. Except for Fig. 4, all charts ŽFigs. 5᎐11. are plotted in logarithmic scale. The values presented are relative ones as compared with a reference material. Thus as you can see, both x- and y-axis are without unit. The properties of the reference material are chosen as the largest values of all SMAs multiplied by 1.5. For example, the largest specific heat is 620 JrKg⬚C ŽNiTi.. Thus, the specific heat of the reference material is 620 = 1.5s 930 JrKg⬚C. We consider three kinds of applied loads, namely, tension, compression and torsion. The parameters in tension are taken as a reference. Note that the combination of material properties, such as specific actuation stress, which is the product of stress over density, may result in a value over 0.66. The following items are identified as key points on selection of SMAs for actuators. 4.1. Transformation temperature and hysteresis Transformation temperature and hysteresis determine the environment in which a SMA may be used. They are critical in a real engineering application. Fig. 4 plots the ranges of each SMA based on the values
presented in Table 1. It is apparent that the transformation temperatures of all three SMAS are close. In terms of hysteresis, NiTi has the broadest range, while CuZnAl and CuAlNi do not share any common range. 4.2. Output work Output work is a key measure of actuators. Within a given space, high output volumetric work Žs stress = strain. ²PI-1: indicates the high output work of an actuator. In the case where lightweight is also important, the PI considered should be output volumetric workrdensity ²PI-2:. Again, high value is preferred. Refer to output volumetric work-density chart ŽFig. 5., the former means that the best choice for output volumetric work is the one that parallels to x-axis and lies towards the top since it corresponds to a high value. For the latter, the one offering the greatest output volumetric work to density ratio parallels to the dash-dotted line and lies towards the upper left corner as it indicates higher value of PI-2. In both one-shot and 100-shot cases, NiTi is the clear winner. Note, N stands for the number of cycle or shot. 4.3. Actuation stress and strain In some applications, one needs high actuation stress. While in others, high strain, which corresponds to large displacement, is required. If lightweight actuator is a goal, instead of actuation stress, specific actuation stress Žs stressrdensity . should be as high as possible. When considering actuation stress and strain, one has to realize the effect of asymmetry in tension and compression in many SMAs. According to w16x, the product of phase transformation stress and maximum
W. Huang r Materials and Design 23 (2002) 11᎐19
15
Fig. 6. Transformation strain vs. actuation stress. Ža. N s 1, tension, Žb. N s 100, tension, Žc. N s 1, compression, Žd. N s 100, compression, Že. N s 1, torsion and Žf. N s 100, torsion.
16
W. Huang r Materials and Design 23 (2002) 11᎐19
Fig. 7. Transformation strain vs. specific actuation stress. Ža. N s 1, tension, Žb. N s 100, tension, Žc. N s 1, compression, Žd. N s 100, compression, Že. N s 1, torsion and Žf. N s 100, torsion.
W. Huang r Materials and Design 23 (2002) 11᎐19
17
Fig. 8. Heating and cooling speed. Ža. Electrical resistivity vs. minimum input power per Kg and Žb. thermal conductivity vs. minimum input power per Kg.
phase transformation strain is approximately constant. Using the scheme proposed in w16x, we can obtain the actuation stress and transformation strain of a given stress state. Fig. 6 is the chart showing transformation strain ²PI-3: y actuation stress ²PI-4: under tension, compression and torsion, respectively. NiTi is the winner in most cases in terms of either large strain or large stress. However, CuAlNi could be the first choice in one-shot, compressive actuator in which large strain is the main goal while actuation stress is less important. Fig. 7 is transformation strain-specific actuation stress ²PI-5: chart, which gives similar conclusion, as that of Fig. 6. 4.4. Heating and cooling speed SMA actuators are actuated by joule heat. As mentioned above, there are three methods to heat up SMA. Actually, they belong to two types. One is by electrical current and the other is by heat transfer. The former depends heavily on electrical resistivity. The latter relates to thermal conductivity. The minimum input power per Kg for actuation includes three terms: output work; latent heat; and thermal energy needed for the increase of temperature in SMA. Heat transfer is ignored in the fast heating case. The exact temperature increase required for actuation depends on the set-up of SMA in terms of pre-strain and stress, hardening in transformation, environmental temperature and the required actuation temperature, etc. In the current study, we assume the required temperature increase is the same as the hysteresis width. If SMA is heated by applying an electrical current, minimum input power per Kgrelectrical resistivity
²PI-6: and minimum input power per Kg= electrical resistivity ²PI-7: are PIs for input current control and input voltage control, respectively. Small value corresponds to high speed. In electrical resistivity y minimum input power per kg chart, Fig. 8a, the former requires the chosen material located towards the lower right corner ŽNiTi is the winner., while the latter asks for the material that lies towards the lower left corner ŽCuZnAl is chosen followed by CuAlNi.. In heating or cooling by means of heat transfer, the heatingrcooling speed within SMA depends on thermal conductivity. PI is minimum input power per Kgrthermal conductivity ²PI-8:. A small value is preferred for high speed. According to thermal conductivity y minimum input power per kg chart, Fig. 8b, CuZnAl Žat the lower right corner. is the best. If low speed actuation is the goal, we should choose in an inverse way as mentioned above. 4.5. Energy efficiency It is necessary to know the energy efficiency of an actuator. In most of the applications, higher energy efficiency is critical. This is described by output work per kgrminimum input power per kg ²PI-9:. The material corresponding to the greatest value has the best energy efficiency. Fig. 9 is output work per kgy minimum input power per kg chart. The winner lies towards the lower right corner. Both NiTi and CuZnAl are more or less at the same level with CuAlNi lagging behind slightly. 4.6. Cost To be of real commercial interest, cost plays a key
W. Huang r Materials and Design 23 (2002) 11᎐19
18
4.7. Damping High damping capacity is an attractive property of SMAs. The comparison among different SMAs should address two aspects: one is damping capacityrcost ²PI-11:; and the other is damping capacityrdensity ²PI-12: or damping capacityrvolume ²PI-13:. The former is to choose the material at the upper left corner of cost-damping capacity chart ŽFig. 10a., which reflects the higher damping capacity at a low cost. For high damping capacityrdensity ratio, in which lightweight is critical in the application, the choice is at the upper left corner of density-damping capacity chart ŽFig. 10b.. If the application is restrained by space limitation, high damping capacityrvolume ratio is important, which is at the upper right corner. It can be seen that CuZnAl is the best choice for damping in all aspects. Fig. 9. Energy efficiency.
5. Conclusions role. Output volumetric workrcost ratio ²PI-10: measures the influence of material cost only. Fig. 10 is cost ratio-output volumetric work chart for both one-shot and 100-shot actuators. The winner should lie towards the upper left corner as it corresponds to low material cost for the same output work. It indicates that CuZnAl is the best, while NiTi is the least. It is necessary to point out that processing cost and running cost are ignored at present. It may be more advantageous to use NiTi because of reduced voltage requirements due to much higher resistivity, which results in cheaper equipment in cyclic applications.
In this paper, a systematic study on the selection of SMAs for actuators is presented. The candidates of current study are NiTi, CuZnAl and CuAlNi, which are the most popular ones at present. More SMAs may be added in to give a complete picture. Thirteen performance indices are identified as key parameters in actuators. The corresponding charts are produced for the selection of SMAs. The current study shows that NiTi is the overall winner with respect to most of the thermo-mechanic related performances. However, when material cost is
Fig. 10. Cost ratio vs. output volumetric work. Ža. N s 1 and Žb. N s 100.
W. Huang r Materials and Design 23 (2002) 11᎐19
19
Fig. 11. Damping. Ža. Cost vs. damping capacity and Žb. density vs. damping capacity.
taken into consideration, CuZnAl should be the first choice, followed by CuAlNi. CuZnAl is the most suitable choice for damping-related actuators. References w1x Duerig TW, Stockel D, Keeley A. Actuator and work production devices. Engineering aspects of shape memory alloy. Butterworth-Heinemann, 1990:181᎐194. w2x Hirose S, Ikuta K, Umetani Y. Development of shape-memory alloy actuators. Performance assessment and introduction of a new composing approach. Adv Robotics 1989;3Ž1.:3᎐16. w3x Otsuka K, Wayman CM. Shape memory materials. Cambridge University Press, 1998. w4x Lee AP, Ciarlo DR, Krulevitch PA, Lehew S, Trevino J, Northrup MA. A practical microgripper by fine alignment, eutectic bonding and SMA actuation. Sens Actuators A 1996;54:755᎐759. w5x Huang W, Toh W. Training two-way shape memory alloy by reheat treatment. J Mater Sci Lett 2000;19:1549᎐1550. w6x Otsuka K. Introduction to the R-phase transition. Engineering aspects of shape memory alloys. Butterworth-Heinemann, 1990:36᎐45. w7x Huang W. Shape memory alloy micro actuators: review and numerical simulation. In: MINDEF-NTU Joint R and D Seminar, 1999, p. 221᎐226.
w8x Bergamasco M, Salsedo F, Dario P. A linear SMA motor as direct-drive robotic actuator, 1989 IEEE International Conference on Robotic and Automation, 1989, p. 618᎐623. w9x Iwanaga H, Tobushi H, Ito H. Basic research on output power characteristics of a shape memory alloy heat engine. JSME Inter J Series I 1988;31Ž3.:634᎐637. w10x Thrasher MA, Shahin AR, Meckl PH, Jones JD. Thermal cycling of shape memory alloy wires using semiconductor heat pump modules. SPIE 1992;1777:197᎐200. w11x Gorbet RB, Russell RA. A novel differential shape memory alloy actuator for position control. Robotica 1995;13:423᎐430. w12x Ditman JB, Bergman LA, Tsao TC. The design of extended bandwidth shape memory alloy actuators, AIAA-94-1757-CP 1994, pp. 210᎐220. w13x Huber JE, Fleck NA, Ashby MF. The selection of mechanical actuators based on performance indices. Proc R Soc: Lond A 1997;453:2185᎐2205. w14x Ryhanen J. Biocompatibility evaluation of nickel-titanium shape memory metal alloy. PhD Dissertation, University of Oulu 1999. w15x Duerig TW, Pelton AR. Ti᎐Ni shape memory alloys, materials properties handbook, titanium alloys. ASM International, 1994. w16x Huang W. ‘Yield’ surfaces of shape memory alloys and their applications. Acta Mater 1999;47Ž9.:2769᎐2776.
On the selection of shape memory alloys for actuators W. HuangU School of Mechanical and Production Engineering, Nanyang Technological Uni¨ ersity, Nanyang A¨ enue, Singapore 639798, Singapore Received 21 June 2000; accepted 2 April 2001
Abstract This paper presents a series of charts for the selection of shape memory alloys ŽSMAs. for actuators. It is based on performance indices with special reference to the unique features of SMA actuators. The three most popular polycrystalline SMAs are candidates of current study, namely, NiTi, CuZnAl and CuAlNi. More SMAs may be added in to give a complete picture. 䊚 2002 Elsevier Science Ltd. All rights reserved. Keywords: Shape memory alloy; Actuator; Selection; Performance index
1. Introduction The unique shape memory behavior in some smart materials attracts a lot of attention from various communities. Among them, shape memory alloys ŽSMAs. show great potential in many applications, in particular, as novel actuators. Many alloys displaying shape memory have been found and considerable effort is still being made to discover new materials. However, the property of these SMAs varies significantly. It turns out to be necessary to have a systematic scheme for the selection of SMAs. There are some preliminary works on this topic w1,2x, which cover some concerns in real applications. The aim of this paper is to present a series of charts for the selection of SMAs for actuators, in particular, at the pre-design stage. Current discussion focuses on the three most popular polycrystalline SMAs, namely, NiTi, CuZnAl and CuAlNi. The discussion is purely based on the material properties reported in the literature. 2. Properties of SMAs A SMA is able to memorize and recover its original U
Tel.: q65-7904859; fax: q65-7911859. E-mail address: [email protected] ŽW. Huang..
shape after being significantly deformed from heating over the phase transformation temperature. This phenomenon is known as shape memory effect, Fig. 1a. On the other hand, at a high temperature, the large deformation can be fully recovered by simply releasing the applied load. This is superelasticity, Fig. 1b. What is involved in both kinds of phenomena is either transformation between low temperature phase Žmartensite . and high temperature phase Žaustenite . or reorientation among different martensite variants. It is well known that SMA is very sensitive to the exact composition, grain size, processing Žincluding heat treatment . and loading conditions w3x. The inclusion of additional elements with minute quantity may change the properties of a SMA dramatically. For instance, the addition of a small amount of copper in NiTi SMA makes the transformation temperature less sensitive to composition and hysteresis gap narrower w4x, adding a small amount of palladium into NiTi SMA, high transformation temperature Žover 200⬚C. is achievable w5x. The sensitivity of SMA seems to be an intrinsic disadvantage. On the other hand, it provides tremendous flexibility to get exactly what is wanted within a far broader range on comparison with traditional materials. Unlike ordinary one-way SMA, which can only remember the high temperature austenite shape, twoway SMA, which can remember both high and low
0261-3069r02r$ - see front matter 䊚 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 2 6 1 - 3 0 6 9 Ž 0 1 . 0 0 0 3 9 - 5
W. Huang r Materials and Design 23 (2002) 11᎐19
12
Fig. 1. Typical behavior of SMA. Ža. Shape memory effect and Žb. superelasticity.
temperature shapes, is obtainable by special training Že.g. w5x.. However, the phase transformation strain is significantly less and their stability and repeatability are uncertain. The R phase is an intermediate between austenite and martensite in some SMAs w6x. Upon cooling NiTi SMA from a high temperature, the material may transform from austenite to R phase first and then to martensite upon further cooling. The hysteresis associated with R phase is normally narrow Ža few degree or less. and the corresponding phase transformation strain upon stressing is small Žapprox. 0.2%.. Therefore, it cannot be used as actuator other than switch.
To date, the work on various SMAs is extensive. However, for most SMAs, the study is far from complete. In view of this, our discussion in this paper focuses on the three most popular SMAs, namely, NiTi, CuZnAl and CuAlNi. Only one-way SMA is considered. Monocrystalline SMAs may have a better thermo-mechanical property than polycrystalline material. Due to the difficulty in fabrication and the very high price, they are not presently of much commercial importance. Hence, only polycrystalline SMA is the interest of the current study. Material properties of each SMA are listed in Table 1. They are mainly taken from w3x and w15x.
Table 1 Main properties of typical SMAsa Žmainly taken from w3x and w15x. NiTi
CuZnAl
CuAlNi
Specific heat ŽJrKg⬚C. Thermal conductivity Ž20⬚C. ŽWrmK. Density ŽKgrm3 . Latent heat ŽJrKg. Electrical resistivity Ž106 ⍀ m. Thermal expansion coefficient Ž10y6 rK.
450᎐620 8.6᎐18 6400᎐6500 19 000᎐32 000 0.5᎐1.1 6.6᎐11
390᎐400 84᎐120 7540᎐8000 7000᎐9000 0.07᎐0.12 17
373᎐574 30᎐75 7100᎐7200 7000᎐9000 0.1᎐0.14 17
Maximum recovery stress ŽMpa. Normal working stress ŽMpa. Fatigue strength ŽNs 106 . ŽMpa. Maximum transformation strain Ž%.
500᎐900 100᎐130 350 6᎐8 6᎐8 2᎐4 Ž3. 0.5 ) 105 28᎐83 y200᎐200
400᎐700 40 270 4᎐6 4
300᎐600 70 350 5᎐6 4
) 104 70᎐100 y200᎐150
) 5 = 103 80᎐100 y200᎐200
2᎐50 400
5᎐20 150
20᎐40 300
Damping capacity ŽSDC%. Grain size Ž m.
15᎐20 1᎐100
30᎐85 50᎐150
10᎐20 25᎐100
Melting, casting and composition control Forming Žrolling, extrusion. Cold-working Machinability Cost ratio
Difficult Difficult Fair Difficult 10᎐100
Fair Easy Restricted Very good 1᎐10
Fair Difficult Very difficult Good 1.5᎐20
Normal number of thermal cycles Young’s Modulus ŽGpa. Shape memory transformation temperature Ž⬚C. Hysteresis Ž⬚C. Maximum overheating temperature Ž⬚C.
a
Additional alloying is not considered.
Ns1 N - 102 N - 105 N - 107
W. Huang r Materials and Design 23 (2002) 11᎐19
13
3. SMA actuators There are three basic types of SMA actuators using one-way SMAs ŽFig. 2.. 䢇
䢇
䢇
Fig. 2a shows a one-way actuator. The SMA element is elongated initially at low temperature and then is heated up to pull element P to the left; Fig. 2b shows a biased actuator, which is capable of moving the element P back and forth by heatingrcooling SMA element; and Fig. 2c shows a two-way actuator, which includes two one-way SMA elements. P may move back and forth by heating and cooling the two SMA elements alternatively.
Basically, SMAs may be heated up by three methods, i.e.: 䢇
䢇
䢇
By passing an electrical current through them. This method is only applicable to small diameter SMA wire or spring; by passing an electrical current through a high resistance wire or tape wrapped around SMA. This method is applicable for bulk SMA; and by hot airrwater or exposing to thermal radiation.
The bandwidth of SMA actuators is usually limited. Cooling speed is the dominating factor. Since small hysteresis gap requires small temperature variation, SMA with small hysteresis is preferred by high bandwidth application. Heat conduction is important in bulk SMA. Experiment shows that 6 ms per cycle can be realized in a switch triggered by 0.1-mm-diameter NiTi wire under free convection condition in air w7x. For larger sized SMAs, additional cooling method is required. Flow liquid can improve the bandwidth dramatically w8x. A fan can provide compressive convection for fast cooling w9x. Other technologies that have been exploited include semiconductors w10x, mobile heat sink w11x and pairs of agonistic᎐antagonistic wires w12x.
Fig. 2. Basic types of SMA actuators using one-way SMAs. Ža. One-way actuator, Žb. biased actuator and Žc. two-way actuator.
Fig. 3. A simple linear SMA actuator. Ža. High temperature and Žb. low temperature.
4. Material selection Actuators offer various performances and operate in many different ways. In terms of materials selection, there are always some special goals to meet. They can normally be presented by means of a performance index ŽPI., while the properties of materials can be presented by a variety of charts w13x. It should be pointed out that not every goal could be represented by means of a PI, in particular for those related to nonthermo-mechanical properties. For instance, in biomedical applications, the only candidate is NiTi due to its excellent biocompatibility w14x, which is lacking of in other SMAs. In practice, one should select materials from the charts according to the particular PI required by a special application. However, since these charts are not available yet, we will have to identify the most important PIs first, produce the corresponding charts and then make the choice.
Fig. 4. Transformation temperature vs. hysteresis.
14
W. Huang r Materials and Design 23 (2002) 11᎐19
Fig. 5. Density vs. output volumetric work. Ža. N s 1 and Žb. N s 100.
Fig. 3 shows the principle of a typical linear actuator. This is the simplest actuator, a weight P hangs at the bottom of a SMA wire. At a low temperature, P moves down Fig. 3a. Upon heating, SMA wire contracts and pulls P up, Fig. 3b. External energy is required only in the heating process. Despite the simplicity, in terms of performance, this actuator represents most of the concerns of SMA actuators. A constant compressive force or a constant torque may replace weight P in a real application. One has to realize the effect of asymmetry in tension and compression in many SMAs. The detail on how to deal with it is presented later. Except for Fig. 4, all charts ŽFigs. 5᎐11. are plotted in logarithmic scale. The values presented are relative ones as compared with a reference material. Thus as you can see, both x- and y-axis are without unit. The properties of the reference material are chosen as the largest values of all SMAs multiplied by 1.5. For example, the largest specific heat is 620 JrKg⬚C ŽNiTi.. Thus, the specific heat of the reference material is 620 = 1.5s 930 JrKg⬚C. We consider three kinds of applied loads, namely, tension, compression and torsion. The parameters in tension are taken as a reference. Note that the combination of material properties, such as specific actuation stress, which is the product of stress over density, may result in a value over 0.66. The following items are identified as key points on selection of SMAs for actuators. 4.1. Transformation temperature and hysteresis Transformation temperature and hysteresis determine the environment in which a SMA may be used. They are critical in a real engineering application. Fig. 4 plots the ranges of each SMA based on the values
presented in Table 1. It is apparent that the transformation temperatures of all three SMAS are close. In terms of hysteresis, NiTi has the broadest range, while CuZnAl and CuAlNi do not share any common range. 4.2. Output work Output work is a key measure of actuators. Within a given space, high output volumetric work Žs stress = strain. ²PI-1: indicates the high output work of an actuator. In the case where lightweight is also important, the PI considered should be output volumetric workrdensity ²PI-2:. Again, high value is preferred. Refer to output volumetric work-density chart ŽFig. 5., the former means that the best choice for output volumetric work is the one that parallels to x-axis and lies towards the top since it corresponds to a high value. For the latter, the one offering the greatest output volumetric work to density ratio parallels to the dash-dotted line and lies towards the upper left corner as it indicates higher value of PI-2. In both one-shot and 100-shot cases, NiTi is the clear winner. Note, N stands for the number of cycle or shot. 4.3. Actuation stress and strain In some applications, one needs high actuation stress. While in others, high strain, which corresponds to large displacement, is required. If lightweight actuator is a goal, instead of actuation stress, specific actuation stress Žs stressrdensity . should be as high as possible. When considering actuation stress and strain, one has to realize the effect of asymmetry in tension and compression in many SMAs. According to w16x, the product of phase transformation stress and maximum
W. Huang r Materials and Design 23 (2002) 11᎐19
15
Fig. 6. Transformation strain vs. actuation stress. Ža. N s 1, tension, Žb. N s 100, tension, Žc. N s 1, compression, Žd. N s 100, compression, Že. N s 1, torsion and Žf. N s 100, torsion.
16
W. Huang r Materials and Design 23 (2002) 11᎐19
Fig. 7. Transformation strain vs. specific actuation stress. Ža. N s 1, tension, Žb. N s 100, tension, Žc. N s 1, compression, Žd. N s 100, compression, Že. N s 1, torsion and Žf. N s 100, torsion.
W. Huang r Materials and Design 23 (2002) 11᎐19
17
Fig. 8. Heating and cooling speed. Ža. Electrical resistivity vs. minimum input power per Kg and Žb. thermal conductivity vs. minimum input power per Kg.
phase transformation strain is approximately constant. Using the scheme proposed in w16x, we can obtain the actuation stress and transformation strain of a given stress state. Fig. 6 is the chart showing transformation strain ²PI-3: y actuation stress ²PI-4: under tension, compression and torsion, respectively. NiTi is the winner in most cases in terms of either large strain or large stress. However, CuAlNi could be the first choice in one-shot, compressive actuator in which large strain is the main goal while actuation stress is less important. Fig. 7 is transformation strain-specific actuation stress ²PI-5: chart, which gives similar conclusion, as that of Fig. 6. 4.4. Heating and cooling speed SMA actuators are actuated by joule heat. As mentioned above, there are three methods to heat up SMA. Actually, they belong to two types. One is by electrical current and the other is by heat transfer. The former depends heavily on electrical resistivity. The latter relates to thermal conductivity. The minimum input power per Kg for actuation includes three terms: output work; latent heat; and thermal energy needed for the increase of temperature in SMA. Heat transfer is ignored in the fast heating case. The exact temperature increase required for actuation depends on the set-up of SMA in terms of pre-strain and stress, hardening in transformation, environmental temperature and the required actuation temperature, etc. In the current study, we assume the required temperature increase is the same as the hysteresis width. If SMA is heated by applying an electrical current, minimum input power per Kgrelectrical resistivity
²PI-6: and minimum input power per Kg= electrical resistivity ²PI-7: are PIs for input current control and input voltage control, respectively. Small value corresponds to high speed. In electrical resistivity y minimum input power per kg chart, Fig. 8a, the former requires the chosen material located towards the lower right corner ŽNiTi is the winner., while the latter asks for the material that lies towards the lower left corner ŽCuZnAl is chosen followed by CuAlNi.. In heating or cooling by means of heat transfer, the heatingrcooling speed within SMA depends on thermal conductivity. PI is minimum input power per Kgrthermal conductivity ²PI-8:. A small value is preferred for high speed. According to thermal conductivity y minimum input power per kg chart, Fig. 8b, CuZnAl Žat the lower right corner. is the best. If low speed actuation is the goal, we should choose in an inverse way as mentioned above. 4.5. Energy efficiency It is necessary to know the energy efficiency of an actuator. In most of the applications, higher energy efficiency is critical. This is described by output work per kgrminimum input power per kg ²PI-9:. The material corresponding to the greatest value has the best energy efficiency. Fig. 9 is output work per kgy minimum input power per kg chart. The winner lies towards the lower right corner. Both NiTi and CuZnAl are more or less at the same level with CuAlNi lagging behind slightly. 4.6. Cost To be of real commercial interest, cost plays a key
W. Huang r Materials and Design 23 (2002) 11᎐19
18
4.7. Damping High damping capacity is an attractive property of SMAs. The comparison among different SMAs should address two aspects: one is damping capacityrcost ²PI-11:; and the other is damping capacityrdensity ²PI-12: or damping capacityrvolume ²PI-13:. The former is to choose the material at the upper left corner of cost-damping capacity chart ŽFig. 10a., which reflects the higher damping capacity at a low cost. For high damping capacityrdensity ratio, in which lightweight is critical in the application, the choice is at the upper left corner of density-damping capacity chart ŽFig. 10b.. If the application is restrained by space limitation, high damping capacityrvolume ratio is important, which is at the upper right corner. It can be seen that CuZnAl is the best choice for damping in all aspects. Fig. 9. Energy efficiency.
5. Conclusions role. Output volumetric workrcost ratio ²PI-10: measures the influence of material cost only. Fig. 10 is cost ratio-output volumetric work chart for both one-shot and 100-shot actuators. The winner should lie towards the upper left corner as it corresponds to low material cost for the same output work. It indicates that CuZnAl is the best, while NiTi is the least. It is necessary to point out that processing cost and running cost are ignored at present. It may be more advantageous to use NiTi because of reduced voltage requirements due to much higher resistivity, which results in cheaper equipment in cyclic applications.
In this paper, a systematic study on the selection of SMAs for actuators is presented. The candidates of current study are NiTi, CuZnAl and CuAlNi, which are the most popular ones at present. More SMAs may be added in to give a complete picture. Thirteen performance indices are identified as key parameters in actuators. The corresponding charts are produced for the selection of SMAs. The current study shows that NiTi is the overall winner with respect to most of the thermo-mechanic related performances. However, when material cost is
Fig. 10. Cost ratio vs. output volumetric work. Ža. N s 1 and Žb. N s 100.
W. Huang r Materials and Design 23 (2002) 11᎐19
19
Fig. 11. Damping. Ža. Cost vs. damping capacity and Žb. density vs. damping capacity.
taken into consideration, CuZnAl should be the first choice, followed by CuAlNi. CuZnAl is the most suitable choice for damping-related actuators. References w1x Duerig TW, Stockel D, Keeley A. Actuator and work production devices. Engineering aspects of shape memory alloy. Butterworth-Heinemann, 1990:181᎐194. w2x Hirose S, Ikuta K, Umetani Y. Development of shape-memory alloy actuators. Performance assessment and introduction of a new composing approach. Adv Robotics 1989;3Ž1.:3᎐16. w3x Otsuka K, Wayman CM. Shape memory materials. Cambridge University Press, 1998. w4x Lee AP, Ciarlo DR, Krulevitch PA, Lehew S, Trevino J, Northrup MA. A practical microgripper by fine alignment, eutectic bonding and SMA actuation. Sens Actuators A 1996;54:755᎐759. w5x Huang W, Toh W. Training two-way shape memory alloy by reheat treatment. J Mater Sci Lett 2000;19:1549᎐1550. w6x Otsuka K. Introduction to the R-phase transition. Engineering aspects of shape memory alloys. Butterworth-Heinemann, 1990:36᎐45. w7x Huang W. Shape memory alloy micro actuators: review and numerical simulation. In: MINDEF-NTU Joint R and D Seminar, 1999, p. 221᎐226.
w8x Bergamasco M, Salsedo F, Dario P. A linear SMA motor as direct-drive robotic actuator, 1989 IEEE International Conference on Robotic and Automation, 1989, p. 618᎐623. w9x Iwanaga H, Tobushi H, Ito H. Basic research on output power characteristics of a shape memory alloy heat engine. JSME Inter J Series I 1988;31Ž3.:634᎐637. w10x Thrasher MA, Shahin AR, Meckl PH, Jones JD. Thermal cycling of shape memory alloy wires using semiconductor heat pump modules. SPIE 1992;1777:197᎐200. w11x Gorbet RB, Russell RA. A novel differential shape memory alloy actuator for position control. Robotica 1995;13:423᎐430. w12x Ditman JB, Bergman LA, Tsao TC. The design of extended bandwidth shape memory alloy actuators, AIAA-94-1757-CP 1994, pp. 210᎐220. w13x Huber JE, Fleck NA, Ashby MF. The selection of mechanical actuators based on performance indices. Proc R Soc: Lond A 1997;453:2185᎐2205. w14x Ryhanen J. Biocompatibility evaluation of nickel-titanium shape memory metal alloy. PhD Dissertation, University of Oulu 1999. w15x Duerig TW, Pelton AR. Ti᎐Ni shape memory alloys, materials properties handbook, titanium alloys. ASM International, 1994. w16x Huang W. ‘Yield’ surfaces of shape memory alloys and their applications. Acta Mater 1999;47Ž9.:2769᎐2776.