Effect of compliant layers within piezoelectric composites on power generation providing electrical stimulation in low frequency applications

Abstract

For patients that use tobacco or have diabetes, bone healing after orthopedic procedures is challenging. Direct current electrical stimulation has shown success clinically to significantly improve bone healing in these difficult-to-fuse populations. Energy harvesting with piezoelectric material has gained popularity in the last decade, but is challenging at low frequencies due to material properties that limit total power generation at these frequencies. Stacked generators have been used to increase power generation at lower voltage levels but have not been widely explored as a load-bearing biomaterial to provide DC stimulation. To match structural compliance levels and increase efficiency of power generation at low frequencies, the effect of compliant layers between piezoelectric discs was investigated. Compliant Layer Adaptive Composite Stacks (CLACS) were manufactured using five PZT discs connected electrically in parallel and stacked mechanically in series with a layer of low modulus epoxy between each disc. The stacks were encapsulated, keeping PZT and overall volume constant. Each stack was electromechanically tested by varying load, frequency, and resistance. As compliant layer thickness increased, power generation increased significantly across all loads, frequencies, and resistances measured. As expected, increase in frequency significantly increased power output for all groups. Similarly, an increase applied peak-to-peak mechanical load also significantly increased power output. The novel use of CLACS for power generation under load and frequencies experienced by typical orthopedic implants could provide an effective method to harvest energy and provide power without the use of a battery in multiple low frequency applications.

1. Introduction

Orthopedic surgeries account for almost 20% of all operating room procedures in the United States and have the highest aggregate hospital costs compared to other specialties, including cardiac and childbirth (Fingar et al., 2014; Weiss et al., 2014). Complete bone healing can be challenging in many orthopedic cases, particularly for diabetic patients and tobacco users (Ho, 2017; Kwiatkowski et al., 1996; Marin et al., 2018; Ricci et al., 2014). Specifically, diabetes and cigarette smoking have been associated with significantly higher rates of pseudoarthrosis, failed fusions, non-unions, and increased time to healing in a variety of orthopedic fracture and fusion procedures (Berman et al., 2017; Browne et al., 2007; Chung et al., 2012; Ganesh et al., 2005; Vo et al., 2011). To overcome these challenges, adjunct therapies are used to enhance bone healing and improve fracture and surgical fusion success rates. Direct current (DC) electrical stimulation has over a 35-year history of successful clinical use in promoting bone healing in multiple orthopedic procedures, including non-unions, spinal fusions, and stress fractures (Brighton et al., 1981, 1975; Haddad et al., 2007; Kane, 1988; Rubinacci et al., 1988). DC stimulation has been linked to osteogenesis (Rubinacci et al., 1988; Yonemori et al., 1996) and Tejano et al. showed significantly improved patient outcomes after spinal fusion with the use of DC stimulation, even without the use of posterior instrumentation (Tejano et al., 1996). Electrical stimulation improves healing rates while keeping cost down and decreasing likelihood of additional surgeries or hospital stays (Khalifeh et al., 2018).

Because of their inherent ability to emit electric charge from mechanical loading, many different types of piezoelectric energy harvesters have been explored as an alternative to batteries (Anton and Sodano, 2007). Several different piezoelectric material configurations have been used to power devices and provide electrical stimulation at the site of desired bone healing (Cochran et al., 1985; Friis et al., 2015; Park et al., 1980, 1981). Park et al. and Cochran et al. utilized monolithic piezoelectric material implanted in vivo and found voltage was generated under physiological loading. However, their results showed no enhanced bone healing through the electrical signals supplied. It was not clear in either study exactly how the electrical signals were conditioned and subsequently delivered in the animal models which could have greatly impacted the material’s ability to produce sufficient current to stimulate bone growth (Cochran et al., 1985; Park et al., 1981).

More recently, Goetzinger et al. utilized stacked piezoelectric element concepts to design a multilayer macro fiber composite spinal fusion implant to generate power and deliver DC stimulation to a titanium electrode at the fusion site (Friis et al., 2015; Goetzinger et al., 2016). Friis et al. studied use of these piezoelectric composites in an interbody fusion implant design in a pilot ovine study. The piezoelectric implants delivered sufficient DC stimulation in sync with mechanical loading to enhance healthy bone growth and lead to successful fusion, as compared to control implants that did not fuse (Friis et al., 2015). In this specific electrical stimulation application, the goal was not to maximize power, but instead to produce sufficient power to induce targeted bone growth at the low frequencies generated by normal human body motion.

While the Friis et al. composite piezoelectric device did deliver sufficient power to improve bone growth, mechanical studies showed peak power output occurred at a range that was much higher than the resistance of a rectifying circuit necessary to convert the voltage to a DC signal (Goetzinger et al., 2016; Tobaben et al., 2015, 2014). Additionally, the difficulty of the fabrication processes associated with the use of fibers prevented scalability for use in medical devices. The use of piezoelectric stacks would improve manufacturability, while lowering source impedance and producing more power at lower resistances, thus overcoming obstacles in practical implementation of the piezo-composites in spinal fusion implants. Additionally, piezocomposite stacks would more closely match the structural compliance of adjacent bone and other biological tissue, as compared to a co-fired stack or monolithic polycrystalline piezoelectric ceramic generator.

The present study investigated the use of Compliant Layer Adaptive Composite Stacks (CLACS) and characterized power at loads representative of low frequency human body motion. It was hypothesized that the inclusion of a compliant polymer layer between piezoelectric discs changes strain patterns within the piezoelectric discs and enhances power production, given constant piezoelectric volume, size, and shape. It was expected that power production would be consistent with results of previous studies on frequency and load under mechanical loading conditions applicable for medical devices.

2. Materials and methods

2.1. CLACS generation

2.1.1. Material considerations

The piezoelectric discs used were commercially available modified Navy Type I Lead Zirconate Titanate, PZT-4 material (SM111, STEMiNC, Doral, FL). This material was chosen for the high coupling coefficient (k_t = 0.45) and desirable mechanical properties appropriate for the physiological loading conditions (STEMiNC). Both the fired-on silver electrodes and electrical poling were completed by the manufacturer under controlled conditions. The discs were axially poled though the thickness (3-direction), with the positive and negative electrodes on the top and bottom faces. Because the discs were loaded axially in compression, the thickness poling direction would best utilize the power generation characteristics of the discs. The matrix material used was a room temperature cure, two-part, medical grade epoxy (EPO-TEK^® 301, Epoxy Technology, Billerica, MA). In its cured state, EPO-TEK 301 has similar mechanical strength properties as common polymers used in implantable medical devices, and has desirable dielectric properties for use in a piezocomposite.

2.1.2. Specimen fabrication

CLACS with three different compliant layer thicknesses (n = 5 in each group) were fabricated using five 10 × 0.4 mm PZT discs and were encapsulated in epoxy, keeping volume of PZT, overall height, and surface area constant (Fig. 1). The five PZT discs were electrically connected in parallel using conductive epoxy (EPO-TEK^® H20E, Epoxy Technology, Billerica, MA) and thin strips of copper foil. The positive poling direction of each disc was verified, and discs were connected in a chain. This connection method was used for its feasibility for modification with the compliant layers and ease of repeatability in a laboratory setting. The chains of PZT discs were folded in an accordion manner to create stacks mechanically in series. This resulted in alternating poling directions of the discs but maintained parallel electrical connection and separation of the positive and negative electrodes. Slices of 11 × 11 mm cured epoxy of varying thicknesses (0.4 and 0.8 mm ± 0.02 mm) were made to create the compliant layers. For the 0.0 mm CLACS group, a minimal amount of epoxy was used to adhere the discs together to create a stack. For the remaining groups, the compliant layers were adhered and interdigitated between the discs to create the CLACS. All stacks were encapsulated with EPO-TEK 301 to create 11 × 11 × 9 mm specimens. The volume of PZT (157 mm³), volume of epoxy (932 mm³), overall height and surface area were kept constant throughout all specimens. Electrical connectivity and system-level impedance was verified before each stack was electromechanically tested.

Fig. 1.

Side view of compliant layer adaptive composite stacks. (a) 0.0 mm CLACS, and (b) 0.8 mm CLACS.

2.2. Electromechanical testing

Specimens were electromechanically tested to compare voltage produced at varying mechanical loads, frequencies, and resistance loads. A 1200 N preload was applied, followed by cyclic compression at three peak-to-peak loads of 100 N, 500 N, and 1000 N at varying frequencies of 1 Hz, 2 Hz, 3 Hz and 5 Hz using an MTS MiniBionix 858 (MTS, Eden Prairie, MN) with a self-aligning platen and 2.5 kN load cell. Conservative estimates for loads seen in generic orthopedic implants were considered to develop the loading conditions and characterize CLACS behavior during typical human body motion (Arshad et al., 2017; Cromwell et al., 1989).

For each loading condition, voltage output of the stack was measured across a shunting resistance sweep ranging from 15 kΩ to 63.4 MΩ. Resistance values were chosen to characterize the behavior of the stacks at lower resistances necessary for circuit design, as well as to capture the resonance behavior at the matched impedance. A sampling rate of 512 Hz was used for all test conditions and data was collected for 15 cycles to capture the steady-state behavior. A custom MATLAB (Mathworks, Natick, MA) code was generated for data analysis. The measured voltage was converted to RMS, V_RMS = V_out/√2, and the average amplitude of the middle 5 cycles was used for power calculations. Power for each loading condition and resistance was calculated, P = V_RMS²/R, and a two-way ANOVA was used to compare power production as a function of compliant layer thickness and resistance for each load. Tukey-Kramer post-hoc analysis was performed to determine differences between groups (α = 0.05). The log transformation of the data was used to satisfy normality and equal variance assumptions.

3. Results

3.1. Power generation

The power generation capability of each CLACS was characterized over a shunting resistance sweep of 15 kΩ to 63.4 MΩ. Fig. 2 shows the average power generated as a function of compliant layer thickness for a 2 Hz, 1000 N sine wave input force, chosen to represent typical loading on an implant while walking. The shape of the power generation curve in Fig. 2 was consistent for all loads and frequencies tested. The addition of a compliant layer did not affect the source impedance, as each stack type exhibited maximum power at the same resistance (6 MΩ) for the 2 Hz frequency. Table 1 shows the effect of the compliant layer on overall maximum power output for each of the CLACS.

Fig. 2.

Average power output as a function of compliant layer thickness and resistance load. Average power generation curve for all groups at 1000 N and 2 Hz loading condition.

Table 1.

Average maximum power output measured with respect to compliant layer thickness (1000 N, 5 Hz, 2.5 MΩ).

Compliant layer thickness (mm)	Average maximum power (μW)
0.0	3036 ± 267
0.4	3912 ± 708
0.8	4883 ± 813

As expected, maximum power generation occurred at the highest tested load and frequency (1000 N, 5 Hz) for all CLACS. Maximum power increased by 29% and 61% for the 0.4 mm and 0.8 mm groups respectively as compared to maximum power from the 0.0 mm baseline (p < .05). Additionally, the 0.8 mm group produced significantly more power than the 0.4 mm group with a 25% increase (p < .05). This relationship held true across all 12 loading conditions tested (p < .05).

3.2. Frequency

The average power generated for each stack type as a function of frequency and compliant layer thickness is shown in Fig. 3. Average power was reported at the resistance corresponding to maximum power for each frequency (12 MΩ at 1 Hz, 6 MΩ for 2 Hz, 4 MΩ for 3 Hz, 2.5 MΩ for 5 Hz). As frequency increased, average power occurred at a lower resistance. The statistical differences between compliant layer groups seen in the power generation curves and overall maximum power were consistent for all frequencies tested. The 0.4 mm and 0.8 mm CLACS produced significantly more power than the 0.0 mm CLACS for all frequencies (p < .05). The 0.8 mm CLACS also produced significantly more power than the 0.4 mm CLACS (p < .05). These results were consistent for all applied loads and all resistances. Additionally, an increase in frequency significantly increased power output for all groups (p < .05). At each given load level tested, the increase in power generation due to the increasing frequency was as expected and primarily linear, with a two-fold increase from 1 to 2 Hz, a three-fold increase from 1 to 3 Hz, and a five-fold increase from 1 to 5 Hz. These relationships were consistent for all CLACS types for all compliant layer thicknesses for the load levels, resistances, and frequencies tested.

Fig. 3.

Average power generation as a function of compliant layer thickness and frequency at 1000 N. Average power presented at the resistance corresponding to peak power for each frequency (12 MΩ for 1 Hz, 6 MΩ for 2 Hz, 4 MΩ for 3 Hz, 2.5 MΩ for 5 Hz). * represents significant difference (p < .05).

3.3. Load

Increased mechanical load levels significantly increased power output for all groups (p < .05). With the increase in load, there was a consistent percent increase in power for all frequencies, resistances, and CLACS type. At a given frequency, the maximum power generated was approximately 100 times greater with an increase from 100 N to 1000 N, approximately 27 times greater with an increase from 100 N to 500 N, and approximately 4 times greater with the increase from 500 N to 1000 N. This nonlinear increase was consistent throughout all specimens and loading conditions. Fig. 4 demonstrates the relationship of average power generation as a function of compliant layer thickness and mechanical load applied at 5 Hz and 2.5 MΩ. At 5 Hz, maximum power generation occurred at 2.5 MΩ for all CLACS types. The statistical relationships between compliant layer groups were the same in varying loads as with frequency variation. Both the 0.4 mm and 0.8 mm compliant layer thickness CLACS produced significantly more power as compared to the 0.0 mm (p < .05), and the power generated from the 0.8 mm CLACS was significantly greater than 0.4 mm CLACS (p < .05).

Fig. 4.

Average power as a function of compliant layer thickness and load at 5 Hz and 2.5 MΩ. * represents significant difference (p < .05).

4. Discussion

This study was designed to investigate the feasibility of CLACS for use in enhancing power generation under human motion loading, specifically for use in implantable orthopedic devices to increase bone healing with DC stimulation. The goal was to measure the effect of compliant layers between PZT discs and quantify increased efficiency in harvesting energy at low frequencies due to increasing compliant layer thickness. The loads and frequencies tested in the present study define the power generation capability of CLACS under conservative estimates of loading in generic orthopedic implants, but there are other off-resonance frequencies and applications (i.e., civil infrastructure) for which CLACS could be used as an efficient energy harvesting mechanism (Roundy et al., 2003).

The results of this study showed that there was a significant increase in power due to compliant layers between PZT discs in a stack at all loads and frequencies, and across all resistances measured. The addition of the compliant layer between each disc increased the positive strain in the in-plane directions of the disc while compressive loads are applied to the stack in the through-thickness direction, thus effectively amplifying the piezoelectric effect of the material and increasing the voltage (and power) produced (Li et al., 2014). Although a different type of PZT composite was used, Cao et al. (1993) and Challagulla and Venkatesh (2009) both showed that the deformation relationship between the piezoelectric ceramic and an elastic material in a 2–2 layer composite predicted an increase in the piezoelectric effect of the active material due to an inhomogeneous shear stress and enhancement of the piezoelectric coupling coefficient. Additionally, Bayrashev et al. (2004) demonstrated that an increase in strain of the PZT increased the generated electric field by 35%, further supporting the notion that increase in strain leads to an increase in power.

It is known that stacks of piezoelectric elements connected electrically in parallel and stacked mechanically in series lowers the source impedance, but to date little has been done to investigate the influence of interdigitated compliant layers in piezoelectric stacks. Platt et al. showed that a stack of PZT elements lowers the source impedance, allowing maximum power generation at a lower load resistance, as compared to a monolithic material of the same volume (Platt et al., 2005). To the best of our knowledge, no other studies have shown that the use of a compliant layer between PZT discs (i.e., CLACS) results in a significant increase in power generated while keeping PZT volume constant. Because the fundamental resonant frequency of most piezoelectric stacks used to convert mechanical energy to usable electrical power is in the kHz or MHz range, harvesting energy at lower frequencies can be very difficult (Zhao and Erturk, 2014). Including a compliant layer within the stacks could increase efficiency of energy harvesting at frequencies seen in human body motion, civil infrastructure systems, and other low frequency applications.

The novel CLACS structure increased efficiency and lowered the source impedance at maximum power while maintaining the tough mechanical properties. Both attributes are required for use in medical implants and devices subjected to relatively low frequency loading conditions (Friis et al., 2015; Goetzinger et al., 2016). While other studies have used human motion to harvest power from piezoelectric materials, it is challenging to directly compare results because of differences in piezoelectric material properties and volume, geometric configurations, and loading conditions, all significantly affect power production. Although geometric configurations and piezoelectric volume and type were not identical to the present study, Goetzinger et al. tested spinal implants under the same loading conditions. At 1000 N and 2 Hz, a loading condition that mimics loads in the spine during walking (Ignasiak et al., 2018), implants used in Goetzinger et al. generated 566 μW at 16 MΩ when converted to RMS. The volume of PZT used in that study was 217 mm³ (Goetzinger et al., 2016). Despite having 27% less volume of PZT (157 mm³), the 0.0 mm, 0.4 mm, and 0.8 mm CLACS, generated 804 μW, 991 μW, 1243 μW respectively under the same loading conditions and applied resistance. Additionally, the maximum power from all CLACS groups occurred at 5 MΩ and produced more power at every applied resistance.

The addition of the compliant layer did not change the resistance at which maximum power was generated for all CLACS groups. It is important that a DC signal is delivered to the desired bone healing site, so the alternating signal produced by the piezoelectric material must be conditioned and rectified. A rectification circuit small enough to fit within an orthopedic implant would also have a small resistance. The ability of the CLACS to produce significantly more power than traditional stacks at lower resistances is beneficial for medical device design.

A lower bound theoretical model for ideal power generation from a stack with no compliant layers (0.0 mm CLACS) was developed. Assuming the PZT discs were loaded in pure compression with uniaxial deformation, the work (W) done on each disc was calculated given the elastic modulus of the PZT material reported by the manufacturer (Y₃₃ = 73 GPa) and 1000 N load applied (STEMiNC). A lower bounding expression for power (P) can be estimated assuming through thickness compression of the piezoelectric elements using the electromechanical coupling coefficient (k₃₃) and the frequency (f) of the applied load as seen in Eq. (1). This relationship stems from the definition of the k₃₃ material property, which is a measure of piezoelectric capacity of the material, and describes the relationship between electrical energy generated per mechanical load applied (Jaffe et al., 1971). The value used in this calculation was the k_t coefficient (0.45) provided by the manufacturer (STEMiNC).

$$P=W*f*k_{33}^2$$ (1)

The results from the present study found that the relationship between power generated and frequency increase is approximately linear. These results are further confirmed by the relationship between power and frequency in the theoretical model (Fig. 5). This same relationship held true as the compliant layer increased in thickness. However, it is not clear if the increase in power due to the addition of compliant layers would also hold true for higher frequencies due to the viscoelastic nature of the compliant material. The theoretical model for pure PZT stacks predicts the maximum power generation (1000 N, 5 Hz) capability of the 0.0 mm specimen without any compliant layers within 30%. As shown in Fig. 5, the 0.0 mm specimens produced more power than expected for all frequencies. This difference could be explained in part by the manner in which the 0.0 mm specimens were manufactured.

Fig. 5.

Comparison of experimental results and theoretical predictions of power generation and the effect of compliant layer thickness with respect to frequency at 1000 N and the resistance corresponding to maximum power.

The discs in the 0.0 mm stacks were adhered together mechanically in series with a thin layer of epoxy to ensure correct alignment once encapsulated. The additional deformation this thin layer of epoxy allowed in the discs in the 0.0 mm CLACS could explain the difference from the theoretical model. The theoretical model under predicted power for the 0.0 mm thickness CLACS, suggesting the small amount of epoxy could have accounted for part of the 30% difference. This suggests that even a small compliant layer could increase power generation as compared to a pure PZT stack. Understanding the fatigue and multiaxial loading behavior of materials used in orthopedic devices is critical to design and mechanical assessment (Friis, 2017). A pure PZT stack has poor fatigue resistance because it is ceramic (Platt et al., 2005). The compliant layer, even a very thin layer, would not only increase power as seen in CLACS, but could also increase the mechanical toughness of the material and improve fatigue behavior.

Future work should include further mechanical testing of the CLACS to characterize the failure of the material in different loading conditions, as necessary for testing of orthopedic implants. Power generation of CLACS was only characterized under pure compressive loads, but applying torsion or multiaxial loads would further explain the power generation capabilities and efficacy for use in implants that experience off axis loading. Additionally, this study only investigated compliant layers that were up to twice the thickness of the PZT discs. Further work could include finding the limit of the increase in power generation due to addition of the compliant layer. Finally, the lower bounding theoretical model used in this study did not take into account the influence of the compliant layer on power generation. Theoretical models are currently being developed to explain the amplification mechanisms due to the compliant layer.

5. Conclusions

The addition of a compliant layer between PZT discs to form CLACS significantly increased the power production capabilities of PZT stacks across all compressive mechanical loading conditions and resistances, while PZT volume remained constant. Previous studies have shown that piezoelectric stacks can produce identical power compared to a monolithic element at a lower impedance, but have not shown an increase in power generation with a constant volume of piezoelectric material as seen in this study. The novel use of compliant layers in piezoelectric stacks for power generation could provide an effective method for energy harvesting, without the use of a battery, in low frequency applications.