Skip to main content

Using an environmentally benign and degradable elastomer in soft robotics


This work introduces an environmentally benign and degradable elastomer, poly(glycerol sebacate) with calcium carbonate (PGS-CaCO\(_{3}\)), for use in soft robotics. Development of greener materials like PGS-CaCO\(_{3}\) contributes to robot designs that do not require retrieval and can safely degrade in the natural environment. A simplified synthesis method of PGS was used to create elastomer sheets, which were laser cut/rastered then laminated with cyanoacrylate glue into pneumatic soft actuators. The modified polymer synthesis method is accessible for roboticists and the three chemicals used are non-hazardous and inexpensive. Three accordion-style pneumatic actuators (3, 4 and 5 chambers) were characterized for free displacement and blocked force in both linear extension and curling motions, and an additional four 3-chambered actuators were also tested to leakage and failure. Material characterization of PGS-CaCO\(_{3}\) samples of all ages gave: ultimate tensile strength (UTS) from 48 to 160 kPa, elongation percent at UTS from 157 to 242%, moduli from 45 to 154 kPa, average resilience of 88% at 100 cycles, and maximum compressive force of 246 N at 50% strain. After being in an approximately 50–55 \(^\circ \)C compost pile for 7 days, the polymer visibly degraded and had an average mass loss of 20% across 12 samples. PGS’s strength, elasticity, biodegradability and chemical safety make it a desirable option for roboticists looking to leverage sustainable materials. PGS may also prove a potential green alternative for robotics applications in ubiquitous environmental and infrastructure sensing.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11


Download references


The authors thank Dr. Skip Rochefort for his guidance regarding polymer processing and chemistry, and Andrew Brickman, Dylan Thrush and John Morrow IV for equipment assistance.

Author information

Authors and Affiliations


Corresponding author

Correspondence to Stephanie Walker.

Ethics declarations

Conflict of interest

No competing financial interests exist.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary material 1 (pdf 14 KB)




Equipment: Regulator: Smith, for Nitrogen tank

Flow Meter: Dwyer RMA-150-SSV

Hot Plate: IKA C-MAG HS 7

PTFE-coated liner: Linden Sweden–Jonas of Sweden

Metal baking pan: Nordic Ware

Vacuum oven: VWR Scientific 1410

Thermocouple Readers: Amprobe TMD-56

Aluminum reaction block: Scilogex

Tensile testing dimensions

Tensile testing dimensions are based on the ASTM D412 Die D dimension restrictions for height 16 ± 1 mm, width 100 mm minimum, gauge length 33 ± 2 mm, and large curvature radius 16 ± 2 mm (Fig. 12). The gauge width was kept wider than the Die D dimension of 3 mm to reduce chances of excess stretching while peeling and handling the samples.

Fig. 12
figure 12

Tensile and cyclic loading were performed with dumbbell shaped PGS-CaCO\(_{3}\) samples of this size. Sheet thickness of PGS-CaCO\(_{3}\) ranged from approximately 2.6–3.8 mm

Leakage and failure

Photos from leakage and failure testing are shown in Fig. 13.

Fig. 13
figure 13

Failure tests for Actuators D, E, F, and G. Extension at failure ranged from 106% at 19.7 kPa to 136% at 17.2 kPa, and failure for force applied ranged from 1.84 N at 12.5 kPa to 0.064 N at 30.8 kPa due to slipping of the curved actuator

Motion capture

Motion capture markers were placed on two of the four chambers of Actuator B, two markers per chamber. The grid was drawn by hand with a permanent marker and a ruler using 0.5 cm spacing. The actuator’s pressure was increased from 0 to 6.9 kPa evenly across 22 s using a syringe controlled by hand. The motion was recorded both by a four-camera OptiTrak system and a Nikon D7000 camera looking down towards the sample. The locations of the markers in the 2D camera image were found automatically; the grid points were located manually on every 30th frame. The 2D data was aligned to the 3D data by solving for the camera location using the known 2D and 3D locations of the OptiTrak markers, and initial depth of the 2D grid by approximating the actuator’s geometry with a cylinder. The 3D data was mapped to a canonical cylinder by finding the rigid body transformation that minimized the difference between the OptiTrak points and canonical matching points on the cylinder.

Bend was calculated by taking the angle between the two markers on the bottom chamber and the two on the top; bend is the angle along the cylinder axis. Extension was calculated by taking the average between each of the points on each chamber and then taking the midpoint between the two separate chambers.

The motion capture data for Actuator B is plotted in Fig. 14. All data is plotted with respect to the pressure. The sharp slope at approximately 1.4 kPa is due to the slippage of the actuator on the table. The elongation of the actuator is slightly nonlinear with increasing pressure, but this could be due to the slipping of the actuator or uneven air flow from the pushed syringe.

Fig. 14
figure 14

Motion captured data of the bend and strain vs. pressure for a four-chambered actuator. The sharp slope at approximately 1.4 kPa is due to the slippage of the actuator on the table

Compression testing

Compression testing was performed on 20 mm diameter circular samples using the Mark-10. Samples were placed on top of sandpaper slightly larger than 20 mm in diameter on top of the bottom compression plate. The top compression plate applied force at a constant rate of 3 mm/min until the center of the sample was reached and then traveled upwards at 3 mm/min to complete one cycle. The first three cycles of compression per sample are plotted.

Compression testing results are shown in the Fig 15. The three samples represent the overall spread of data. The samples overlapped and reached force values from 170 to 246 N at 50% strain.

Fig. 15
figure 15

Compression testing for three samples of different processing conditions and ages. Each sample underwent three compressive cycles at 3 mm/min to 50% of the polymer height. These three samples represent the overall spread of data, and reached from 170 to 246 N

CO\(_{2}\) Calculation

Calculation of CO\(_{2}\) release should all CaCO\(_{3}\) dissociate:

Molar mass CaCO\(_{3}\) = 100.0869 g/mol

Ideal mass of CaCO\(_{3}\) going into melt = 0.476 g

0.00476 moles of CaCO\(_{3}\) in = 0.00476 moles of CO\(_{2}\) out = 0.2095 g CO\(_{2}\) out

“19.64 pounds of carbon dioxide (CO\(_{2}\)) are produced from burning a gallon of gasoline that does not contain ethanol”


19.64 lbs CO\(_{2}\)/gal burned = 8909 g CO\(_{2}\)/gal burned

Assuming your car gets 30 miles/gallon, burn 297 g/mile

297 g/mile / 5280 ft/mile = 0.05625 g/ft

0.2095 g CO\(_{2}\) from polymer/0.05625 g/ft = 3.724 ft

1000 sheets of polymer (approximately 1000 small actuators) = 3724 ft = 0.705 miles

Preparation of natural latex rubber and silicone tensile samples

Natural rubber centrifuged latex (in water) from Liquid Latex Fashions (clear) was poured into a sheet mold and left to dry for 5 days. The rubber sheet was then laser cut with the same laser settings as PGS-CaCO\(_{3}\) and in the same size dumbbells as the PGS-CaCO\(_{3}\). The rubber sheet thicknesses ranged from 1.4 to 1.7 mm.

50 wt% Part A and 50 wt% Part B of Ecoflex 00-30 were mixed at 2000 rpm for 30 s and then 2200 rpm for 30 s in a Thinky. The mixture was then poured into the same dumbbell shaped molds as the rubber samples (after mold cleaning) and intermittently vacuumed for 15 min in a vacuum chamber to remove any bubbles. The samples were then placed in a 60 \(^\circ \)C oven for 20 min. After being taken out of the oven, sandpaper was glued to the edges of each dumbbell on each side to reduce slipping during testing. The Ecoflex dumbbell thicknesses ranged from 2.3 to 2.9 mm. There was a slight lip on the outer edges of the dumbbell where the silicone met the acrylic mold, so dumbbell thickness was determined using the center of each sample.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Walker, S., Rueben, J., Volkenburg, T.V. et al. Using an environmentally benign and degradable elastomer in soft robotics. Int J Intell Robot Appl 1, 124–142 (2017).

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI:


  • Elastomer
  • Green chemistry
  • Soft robotics
  • Actuator
  • Degradable