Keywords

1 Introduction

The manipulation of objects or components at the micrometer scale is a challenging task. As objects become smaller, disturbances come into play that can usually be ignored for the manipulation of objects with a larger size. Examples include effects due to surface interaction (e.g. Van der Waals force and electrostatics [1]) and the dominance of viscous forces when the surrounding environment is fluidic (e.g. water [2, 3]). Most commonly, locomotion of such untethered microdevices is either executed on a planar surface or in full 3D. The design (i.e. size, shape and actuation mechanism) of the device therefore prescribes its motion capabilities and the disturbances that will act upon it. When the size of the end-effector has the same order of magnitude as the size of the object, in general, manipulation becomes easier. Considering the microrobot as an end-effector, object manipulation differs greatly compared to manipulation with larger tools such as robotic tweezers or probes. Such tweezers are usually inserted from a certain angle and are limited in dexterity and degrees of freedom for manipulation. Moreover, when the size of the microrobot decreases, so does the force necessary for actuation. For instance, the drag force acting on a sphere in viscous flow scales with R, the radius of the sphere, while the gravitational force scales with the volume of the sphere, and thus with R 3. Additionally, when the object scales down such that the Reynolds number Re ≪ 1, the motion of the microrobot and of objects will be more damped, and local fluidic effects can be exploited. In particular this includes the creation of a vortex due to the local rotation of the microrobot and a fluid’s no-slip condition on the surface. Both can be used to aid in the lifting and/or transportation of objects. All these characteristics enable actuation techniques unconventional for macro-scale robotics and include magnetic, optical and fluidic techniques.

1.1 Related Work

Many different tools, designs and actuation techniques exist for micro- and nanomanipulation (see, e.g., [4]). A comparison regarding magnetically actuated microrobots can be found in [5, 6]. Moreover, a review on methods in automated planning and control for micromanipulation, and contact micromanipulation in particular can be found in [7, 8], respectively. Propulsion mechanisms with magnetic fields can be classified by the way energy is transferred to the robot, either force based with magnetic field gradients or torque based with alignment to a magnetic field. Examples of magnetic gradient control can be found in [5, 9, 10], while torque-based actuation techniques can be found in [3, 11, 12].

Regarding the manipulation of objects, research and applications in the microscale are single cell studies [13], minimally invasive surgery [14] and micro-assembly [15]. Other work considering manipulation using an untethered microrobot can be found in [12, 16]. Automated manipulation via fluidic trapping (i.e. pick-and-place) of beads with a similar microrobot as presented in this work is demonstrated in [17]. Additional work considering similar automated fluidic trapping strategies can be found in [16].

1.2 Micromanipulation Environment

Microrobot and micro-object manipulation is often executed in a low Reynolds number fluid environment. In this regime viscous forces are dominant and have to be overcome to achieve motion. Particular details of fluid mechanics at low Reynolds number can be found in many textbooks [18]. Other work, describing life at low Reynolds number and robotics at the microscale can be found in [1, 2], respectively. One important effect that a low Reynolds number fluid has on bodies is the considerable influence of drag. When in motion, a translating and rotating body will be disturbed by a drag force and drag torque, respectively. This drag force and torque increases with increasing linear and rotational velocity of the body and with the viscosity of the surrounding medium. Other effects that occur due to the size scale are surface interaction forces such as adhesion or Van der Waals forces [19].

One additional property of a low Reynolds number flow that is worth mentioning is its reversibility. When Re ≪ 1, meaning a laminar and reversible flow, the microrobot can move forwards and backwards and have an almost instantaneous flow reversal. This means that when an object is transported without contact in the vortex, a sudden motion reversal of the robot will not result in the loss of the object.

In this chapter we present a family of microrobots that are actuated via torque with a rotating and stepping magnetic field. Additionally, we show the object manipulation capabilities of these microrobot designs. The rotating magnetic field enables a fluidic-based noncontact manipulation technique, while a stepping magnetic field enables a contact manipulation technique. We emphasize that these different robotic designs result in different actuation techniques and object manipulation approaches, while the fabrication process remains the same. By altering the internal and external configuration of the microrobot, that is, by changing the size, shape and configuration of the internal soft-magnetic posts with respect to the microrobot’s body, entirely different propulsion and manipulation mechanisms are possible.

2 Microrobot Design

The microrobot (see Fig. 4.1 and Table 4.1) is a wireless mobile device and has a width of 60–\(200\,\upmu\) m and a length of 150–\(600\,\upmu\) m. It is designed as a cuboid (SU-8) with internal soft-magnetic posts consisting of a cobalt–nickel (CoNi) alloy that aligns with an external magnetic field. When the field direction changes, the microrobot realigns with it [20].

Fig. 4.1
figure 1

Top view of the microrobot design. The geometry of the microrobot is defined by the length l and width w of the body. The configuration of the soft-magnetic posts differs in pitch between the posts d p and the number of posts n p . All cylindrical soft-magnetic posts have a diameter \(D = 30\,\upmu\) m and a height of \(h = 50\,\upmu\) m. The grey area depicts CoNi, the white area depicts SU-8. The type description of the device is explained in Table 4.1

Full size image
Table 4.1 Microrobot design configuration
Full size table

2.1 Design and Capabilities

To give the microrobot as much torque as possible for actuation different configurations were designed. These designs differ from each other in geometry, number and shape of soft-magnetic posts and their configuration in the device. Geometrically, these microrobots were fabricated with different length l and width w of the body. The configuration of the soft-magnetic posts varies in pitch between the posts d p , the number of posts n p as well as the shape (cylindrical or rectangular). All cylindrical soft-magnetic posts have a diameter \(D = 30\,\upmu\) m and a height of \(h = 50\,\upmu\) m. Several configurations are detailed in Fig. 4.1 and their type name is explained in Table 4.1. For example, a microrobot with length \(l = 300\,\upmu\) m, width \(w = 60\,\upmu\) m, number of posts n p  = 5, spaced d p  = 2D apart, would be denoted as RL3N5.

According to [17], the torque can be increased with either a stronger magnetic field H, a more pronounced easy axis, or with increasing volume V of the magnetic material. The limitation of increasing magnetic material is related to the preferred magnetization direction (easy axis) of the device. In order to achieve the desired rolling motion (left two designs in Fig. 4.1) the microrobot has to be kept as symmetrical as possible along its long axis and has a transverse magnetization. This is achieved with the internal soft-magnetic posts. When the pitch between the soft-magnetic posts d p is too small, the magnetization direction can change. The ideal configuration, therefore, depends on d p , and defines the number of posts n p . For increasing pitch distance d p the soft-magnetic posts do not influence each other and behave like individual cylinders. The minimal configuration was experimentally found with d p  = 2D. The dimensions of the soft-magnetic posts are therefore preferred with a diameter of \(D = 30\,\upmu\) m and a height of \(h = 50\,\upmu\) m, thus having a height to diameter ratio of h∕D ≈ 1. 67. This leads to a volume fraction of CoNi in the device up to 20 % (nearly 65 % by weight) allowing a high magnetization and generating sufficient torque to drive the agent in liquid environments, including in higher viscosity fluids.

Regarding the stepping actuation technique of the device (right design in Fig. 4.1), the microrobot is designed as a rectangular cuboid, with internal rectangular soft-magnetic posts. This leads to a height to diameter ratio of h∕D ≈ 2. 92 offering sufficient torque to enable the stepping actuation technique. For both designs, motion is induced due to the friction of the microrobot with the supporting surface, which is explained in more detail in Sect. 4.3.

2.2 Fabrication

For the fabrication of the microrobot, a silicon wafer covered with a silicon dioxide (SiO2) sacrificial layer is used as the substrate, followed by an evaporated titanium (Ti) and gold (Au) bilayer that is patterned by a lift-off process. A thin layer of adhesion promoter, Omnicoat, and a \(50\,\upmu\) m thick layer of SU-8 are spincoated on the wafer and patterned into tethered structures with holes within the bodies. The holes are then filled with CoNi alloy by an electroplating process, followed by another thin layer of electroplated gold to protect the CoNi alloy during the subsequent etching step. The wafer is then diced into small chips and immersed in a buffered hydrofluoric (BHF) acid solution to etch away SiO2 and release the structures from the substrate. Finally, a micro-laser milling machine is utilized to cut the microrobots from their tethers. For more details on the fabrication process, we refer to [21]. This fabrication procedure lends itself perfectly for mass-production of the microrobot and in one batch (wafer) up to 40,000 devices can be fabricated.

Due to its exceptional magnetic properties, electrodeposited soft-magnetic cobalt-nickel (CoNi) is used. The content ratio of cobalt and nickel specifies the magnetic properties of the material. This leads to tunable magnetic properties, ranging from semi-hard to very soft [22]. The used CoNi alloy contains approximately 50 % Co and has high magnetic permeability, high magnetic saturation and low remanence [a saturation magnetization of 0.85 T, and a coercivity of 20 Oe (0.002 T)]. The high magnetic permeability allows the magnetic material to generate higher force or torque in a stronger external magnetic field. A low remanence reduces the residual magnetic force when the magnetic field is turned OFF?

3 Magnetic Actuation

Actuation with magnetic fields is a convenient method for applying power to a microrobot and control its motion. Considering a magnetic field H that is applied in the workspace, the magnetic torque and force that is acting on the microrobot can be defined as

$$\displaystyle{ \boldsymbol{\tau }= V \mathbf{M} \times \mathbf{B} }$$
(4.1)

where M is the volume magnetization in [A∕m] of the object of volume V, the magnetic field has a flux density B = μ 0 H and μ 0 is the permeability of vacuum. The force acting on the microrobot is described as

$$\displaystyle{ \mathbf{F} = V (\mathbf{M} \cdot \nabla )\mathbf{B}. }$$
(4.2)

For the designs presented in this work, the device is actuated by magnetic torque.

3.1 Experimental System

The magnetic field generator (MFG) has eight electromagnetic coils as described in [23] and is commercialized by MagnetbotiX AGFootnote 1. The system is controlled by a single computer and is capable of 5-DOF wireless control of micro- and nano-structures (3-DOF position, 2-DOF pointing orientation) within a spherical workspace with a diameter of approximately 10 mm. This allows for magnetic fields and field gradients up to 20 mT and 2 T/m. The MFG is positioned under a microscope (Olympus IX-81) to observe a container enclosing the microrobot (see Fig. 4.2). For all experiments the magnitude of the magnetic field is set to 10 mT and, as the motion of all presented devices is torque induced, no magnetic field gradient is applied.

Fig. 4.2
figure 2

The experimental system contains the eight-coil magnetic field generator (MFG, white box), a microscope (Olympus IX-81) and a laptop computer for user interaction. Image courtesy of MagnebotiX AG

Full size image

3.2 Rotational Magnetic Fields

Figure 4.3 describes the motion of a rolling microrobot actuated with a rotating magnetic field. If the microrobot rolls without slipping, the maximum forward velocity can be determined as \(\hat{v}_{\mathrm{com},r} = 2 \cdot (w + h)\cdot \omega\), with ω the rotational frequency of the magnetic field. This assumes that the robot follows the field exactly. The forward body length of the microrobot is \(w = 60\,\upmu\) m, resulting in a forward velocity of around 16 body lengths/s at a rotation frequency of 5 Hz. The deviation of the experimental data from this ideal forward velocity indicates slip of the microrobot, as will be explained in Sect. 4.3.4.

Fig. 4.3
figure 3

Side view of a microrobot capable of rolling motion. Applying a rotational magnetic field B enables the microrobot to roll on a supporting surface with a rotational velocity ω and results in a forward motion v com

Full size image

3.3 Stepping Magnetic Fields

Figures 4.4 and 4.5 show the motion of a microrobot of type RW2L3N3 actuated with a stepping magnetic field. If the microrobot steps without slipping, the maximum forward velocity can be determined as \(\hat{v}_{\mathrm{com},s} = l\sin (\phi )\cdot \eta\), with η the stepping frequency of the magnetic field. The forward body length of the microrobot is \(w = 200\,\upmu\) m, resulting in a forward velocity of around 2.5 body lengths/s at a stepping frequency of 5 Hz. The forward velocity does not depend on the tilt angle β of the magnetic field, and was set to β = 40 ∘. The deviation of the experimental data from this ideal forward velocity indicates slip of the microrobot, as will be explained in the next section.

Fig. 4.4
figure 4

Top view (left) and side view (right) of a microrobot capable of stepping motion. Applying a stepping magnetic field B enables the microrobot to tilt and slide on a supporting surface and results in a forward motion v com. \(\varphi\) represents the stepping angle and β represents the tilt angle of the magnetic field

Full size image
Fig. 4.5
figure 5

Snapshots of microrobot actuation via a stepping magnetic field. The microrobot of type RW2L3N3 (\(l = 300\,\upmu\) m, \(w = 200\,\upmu\) m) moves from left to right and covers a distance of approximately 1. 6 mm in the images sequence. This results in a forward speed of around 2. 5 body lengths/s. The black scale bar is \(500\,\upmu\) m long

Full size image

3.4 Evaluation

Figure 4.6 shows the theoretical (dashed lines) and real (solid lines) forward speed for the rolling (red lines) and stepping (blue lines) locomotion technique. For the rolling experiments a microrobot of type RL6N10 is used, while for the stepping experiments a microrobot of type RW2L3N3 is used. The stepping locomotion is evaluated with a stepping angle of ϕ = 45 ∘ and a tilt angle of the magnetic field β = 40 ∘. Because the experimental velocity is lower than the theoretical speed, slip of the microrobot limits the forward speed. One reason for this slip is the viscosity of the surrounding liquid, i.e. when the viscosity is higher this leads to more slip and a lower velocity [20]. Additionally, as forward motion is achieved due to the frictional contact between the microrobot and the surface, the roughness of the supporting surface plays a role as well. When the surface roughness is high, the microrobot is expected to have more traction than for a smooth surface, and the resulting forward speed will be higher. These explanations hold for both the rolling and stepping actuation technique. The motion of the microrobots was evaluated in water (rolling) and isopropanol (stepping) on a glass substrate.

Fig. 4.6
figure 6

Theoretical (dashed lines) and real (solid lines) forward speed for the rolling (red lines) and stepping (blue lines) locomotion technique. The real speed is obtained with water (rolling) and isopropanol (stepping) as liquid environment. For the stepping locomotion technique the microrobot’s length is \(l = 300\,\upmu\) m, the stepping angle ϕ = 45 ∘ and the tilt angle of the magnetic field β = 40 ∘. Each measurement is averaged over three experiments and the error bar represents one standard deviation

Full size image

4 Object Manipulation

The properties of the environment prescribe which object manipulation techniques are possible. Considering a liquid environment, the density and viscosity of the liquid, as well as the density of the object, dictate whether an object can be lifted or not. For example, the density of polystyrene (1. 05 kg m−3) is higher than the density of isopropanol (0. 79 kg m−3, at room temperature). This ensures a polystyrene bead will sink in isopropanol due to gravity, but can still be lifted up due to a generated fluid flow. This delicate balance between the properties of the object and the surrounding fluid demonstrates the importance of environmental conditions and enables two manipulation capabilities. First, manipulation via direct contact is shown, where the microrobot pushes an object. Second, the motion of the rolling microrobot can generate a local fluidic vortex which enables a noncontact manipulation technique.

4.1 Contact Manipulation

Object manipulation by pushing is a common strategy for microrobots and is possible when the object to be manipulated is heavy enough as not to be lifted by the fluid flow and light enough such that the microrobot can actually move it. This means that the apparent weight of the object, and the electrostatic and the Van der Waals forces due to the interaction of the object with the surface, must be lower than the force applied by pushing. Figure 4.7 shows a microrobot of type RW2L3N3 (\(300\,\upmu\) m wide) pushing a micro-object (\(200 \times 350\,\upmu\) m) that is triangular shaped and is fabricated from SU-8 (density 1. 2 kg m−3). Figure 4.8 shows a microrobot of type RL3N5 pushing four micro-objects (SU-8 triangles, \(200 \times 350\,\upmu\) m) such that they densely assemble into a narrow channel (\(750\,\upmu\) m wide). Pushing manipulation is more effective with a rolling microrobot than with a stepping microrobot. This is because the stepping actuation technique achieves less stable motion than the rolling technique, and the orientation of the device can be better controlled with rolling. The orientation of the stepping microrobot is continuously stepping in order to move forward which works against a continuous and stable pushing contact.

Fig. 4.7
figure 7

Microscopic view of object manipulation via pushing. The microrobot of type RW2L3N3 (\(300\,\upmu\) m wide) pushes a micro-object (SU-8 triangle, \(200 \times 350\,\upmu\) m). The black scale bar is \(500\,\upmu\) m long

Full size image
Fig. 4.8
figure 8

Microscopic view of object manipulation via pushing. The microrobot of type RL3N5 (\(300\,\upmu\) m wide) assembles four micro-objects (SU-8 triangles, \(200 \times 350\,\upmu\) m) densely packed into a narrow channel (\(750\,\upmu\) m wide) within 2 min. The black scale bar is \(500\,\upmu\) m long

Full size image

4.2 Noncontact Manipulation

The lifting effect of the fluid flow in front of the microrobot and near the surface occurs due to the no-slip condition of the fluid at the surface (shear flow) and causes objects to be lifted up from the surface. This fluidic lifting effect is local, as objects further away from the robot are not affected. The fluidic vortex above the microrobot is generated due to the robot’s rotation and the fluid’s no-slip condition at the surface. Both effects complement each other, and are possible due to the low Reynolds number environment [24]. Figure 4.9 shows a microrobot of type RL6N10 (\(600\,\upmu\) m wide) lifting and transporting a polystyrene bead that is \(100\,\upmu\) m in diameter.

Fig. 4.9
figure 9

Microscopic view of object manipulation via fluidic trapping. The microrobot of type RL6N10 (\(600\,\upmu\) m wide) approaches (left), lifts (middle) and transports (right) a polystyrene bead with \(100\,\upmu\) m in diameter. The black scale bar is \(300\,\upmu\) m long

Full size image

5 Conclusions

This chapter presented the design and capabilities of a family of microrobots. While the fabrication process is identical for all designs, the method for actuation and the object manipulation capabilities are specific for each particular design. Each microrobot design has embedded soft-magnetic posts that align to a magnetic field. Depending on the arrangement of these posts the microrobot can be actuated by either a rotating magnetic field or a stepping magnetic field. Consequently, each actuation technique enables a different object manipulation techniques (i.e. contact or noncontact manipulation). In particular, a rotating magnetic field enables a local vortex above the robot that traps objects for transportation. Moreover, a pushing manipulation technique can be achieved for both the rotating and stepping magnetic field actuation. Experimental results demonstrate all mentioned techniques for different microrobotic designs.