Page 1
In the format provided by the authors and unedited.
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
NATURE MATERIALS | www.nature.com/naturematerials 1
S1
Supplementary Information
Control of piezoelectricity in amino acids by
supramolecular packing
Page 2
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 2
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S2
Control of piezoelectricity in amino acids by supramolecular packing Sarah Guerin, Aimee Stapleton, Drahomir Chovan, Rabah Mouras, Matthew Gleeson, Cian McKeown, Mohamed Radzi Noor, Christophe Silien, Fernando M. F. Rhen, Andrei L. Kholkin, Ning Liu, Tewfik Soulimane, Syed A. M. Tofail, Damien Thompson Contents Page Figure S1: Dispersion-correction induced external pressure S3 Figure S2: XRD Data S4 Figure S3: Bulk Polymorphic Transformations S5 Figure S4: SEM Image, β-glycine S6 Figure S5: SEM Image, γ-glycine S7 Figure S6: Piezometer Measurement S8 Figure S7: Crystal Slicing S9 Figure S8: Crystal/Probe Alignment S10 Figure S9: Monoclinic Amino Acids S11 Figure S10: Dispersion Corrected DFT Graphs S12 Figure S11: Dispersion Corrected DFT Graphs S13 Figure S12: DFT Energy Convergence S14 Figure S13: DFT Parameter Convergence S15 Figure S14: Piezoelectric Constant Schematic S16 Figure S15: β-glycine grown on PMMA S17
Table S1: DFT Benchmarking S18 Table S2: DFT Benchmarking Percentage Errors S19 Table S3: Physical Constants, L-Asparagine S20 Table S4: Physical Constants, L-Aspartate S21 Table S5: Physical Constants, L-Methionine S22 Table S6: Physical Constants, L-Cysteine S23 Table S7: Physical Constants, L-Isoleucine S24 Table S8: Physical Constants, L-Histidine S25 Table S9: Physical Constants, L-Leucine S26 Table S10: Piezoelectric Matrices, γ-glycine S27 Table S11: Piezoelectric Matrices, β-glycine S28 Table S12: Packing Densities, Glycine Polymorphs S29 Table S13: Dispersion Corrections, Unit Cell Parameters S30 Table S14: Dispersion Corrections, Elastic Constants S31 Table S15: Young’s Moduli Approximations, Glycine Polymorphs S32 Table S16: Piezoelectric Constants, Finite Differences S33
Note: Computational Benchmarking S34 Note: Effectiveness of DFT Predictions S36 Note: Example d16 Calculation from Resonance Data S37 Note: Exploiting Piezoelectricity in β-glycine S40 Note: Experimental Precautions for β-glycine Characterisation S42 Note: Grimme DFT-D3 Dispersion Corrections S44 References S45
Page 3
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 3
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S3
Supplementary Figure 1: Plot of external pressure on the theoretical unit cell with
increasing van der Waals cut off radius. At the dispersion convergence point of 12Å, both β
and γ-glycine have begun to transform into δ and ϵ glycine respectively (Reference 24 in the
main text).
Page 4
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 4
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S4
Supplementary Figure 2: XRD spectrum for evaporated glycine droplets, showing a
mixture of alpha and beta polymorphs (as imaged in main text Fig. 3c). β-glycine needles
could be extracted from the centre of the droplet based on visual inspection. Relative
intensity is in arbitrary units.
Page 5
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 5
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S5
Supplementary Figure 3: Bulk polymorphic transformations during β-glycine
crystallisation. a. α-glycine crystals cluster at the edges of glycine droplets and at the roots of
glycine needles within 24 hours of dropping an aqueous glycine solution. The red, green and
blue dots correspond to the red, green and blue spectra in the following panel. b. Raman
spectra showing the presence of α-glycine at three different points along the crystal face
shown in panel a. c. Over time, or if poorly grown, β-glycine needles become consumed by
their own transformation to the γ or α polymorph. The yellow, blue and pink dots correspond
to the yellow, blue and pink spectra in the following panel. d. Raman spectra showing the
presence of γ-glycine at three different points along the crystal face shown in panel b. The
crystal had been sliced from its substrate approximately 6 hours prior to measurement.
a b
c d
S5
Supplementary Figure 3: Bulk polymorphic transformations during β-glycine
crystallisation. a. α-glycine crystals cluster at the edges of glycine droplets and at the roots of
glycine needles within 24 hours of dropping an aqueous glycine solution. The red, green and
blue dots correspond to the red, green and blue spectra in the following panel. b. Raman
spectra showing the presence of α-glycine at three different points along the crystal face
shown in panel a. c. Over time, or if poorly grown, β-glycine needles become consumed by
their own transformation to the γ or α polymorph. The yellow, blue and pink dots correspond
to the yellow, blue and pink spectra in the following panel. d. Raman spectra showing the
presence of γ-glycine at three different points along the crystal face shown in panel b. The
crystal had been sliced from its substrate approximately 6 hours prior to measurement.
a b
c d
Page 6
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 6
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S6
Supplementary Figure 4: Morphologies of the glycine polymorphs as seen in this Scanning
Electron Micrograph (SEM) for a cleaved β-glycine needle, lying at ninety degrees to the
plane of growth. Needles were sliced to make samples for transverse shear resonance
measurements.
Page 7
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 7
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S7
Supplementary Figure 5: Morphologies of the glycine polymorphs as seen in this Scanning
Electron Micrograph (SEM) for a γ-glycine seed crystal used for energy harvesting
experiments (main text Figure 4).
Page 8
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 8
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S8
Supplementary Figure 6: Quasi-static direct measurement of longitudinal piezoelectric
coefficient using a commercial Berlincourt type piezometer. The figure shows an example
measurement of the d11 coefficient of a γ-glycine crystal that is clamped between two silver
electrodes for simultaneous straining and electrical measurement. The meter (left) reads a
value of -1.7 pC/N. Our DFT calculations predicted a value of-1.6 pC/N for d11 in γ-glycine
(Figure 2b in the main text). Note the instrument readout is labelled d33 because the d11
coefficient is measured by rotating the crystals so that their crystallographic a-axis is in line
with the 3 axis of the meter.
a b
Page 9
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 9
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S9
Supplementary Figure 7: Schematic (left) and corresponding optical images (right) of
monoclinic β glycine crystals illustrating the crystal cut required for piezoelectrically coupled
electromechanical resonance to take place in the thickness shear mode. The ratio of the
thickness, width and length in the samples is around 1:12:30 and a typical slice measures
~200 µm width, ~700 µm length and <20 µm thickness. Microscopic slices from a ~6000 µm
long needle provide samples for the resonance piezoelectricity measurements shown in
Supplementary Figure 9.
Page 10
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 10
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S10
Supplementary Figure 8: Photograph of four microprobes with 3.5 micron radius tips in
contact with a sliced β-glycine needle, as viewed on an ESP MON21 CCTV monitor
connected to a Costar SI-C400N camera.
Page 11
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 11
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S11
Supplementary Figure 9: The seven other smallest (non-glycine) monoclinic L-amino acid
crystals, showing their crystal packing, monoclinic angle, and highest predicted piezoelectric
strain constant. Carbon atoms are coloured grey, hydrogen atoms are white, oxygen atoms are
red, nitrogen atoms are blue and sulphur atoms are yellow.
Page 12
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 12
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S12
Supplementary Figure 10: Convergence in calculated energy, volume, and lattice parameter
values with respect to the vdW cut-off radius used for dispersion corrected DFT calculations,
for β-glycine, γ-glycine, and aluminium nitride (see Supplementary Note p S45). Literature
experimental crystal volumes and lattice constants are from refs 20 and 21 in the main text
and supplementary reference 28.
Page 13
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 13
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S13
Supplementary Figure 11: Computed changes in predicted elastic constant values with
increasing vdW cut-off radius for β-glycine, γ-glycine and aluminium nitride (see
Supplementary Note p. S45).
Page 14
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 14
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S14
Supplementary Figure 12: Convergence of the calculated DFT energies of the α, β, and γ
glycine crystals respectively with respect to k-point sampling. The energies were obtained
from ionic relaxations performed using 1x1x1 to 8x8x8 k-point grids.
Page 15
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 15
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S15
Supplementary Figure 13: Graphs showing the convergence in predicted (top) piezoelectric
and (bottom) elastic constant values for (left) beta and (right) gamma glycine with respect to
k-point sampling.
Page 16
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 16
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S16
Supplementary Figure 14: Relationship between longitudinal (σ) and shear (τ) stress and
corresponding piezoelectric polarisation (P) acting along arbitrarily chosen mutually
orthogonal 1-2-3 axes for four piezoelectric strain constants, d22, d33, d16 and d14. Grey
shading indicates the corresponding surfaces on which the electrodes are contacted to
measure the polarisation.
Page 17
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 17
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S17
Supplementary Figure 15: A selection of images showing β-glycine needles grown on a
layer of PMMA. The PMMA was spincoated onto Kapton film and sprayed with ethanol1,2 to
encourage the growth of the β-polymorph.
Page 18
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 18
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S18
Supplementary Table 1: Calculated piezoelectric and elastic constants for aluminium
nitride, zinc oxide and α-quartz. Other theoretical and experimental results are shown for
comparison3-9. In cases where many studies exist, the values shown are averages across all
studies.
Page 19
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 19
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S19
Supplementary Table 2: Percentage errors in physical constants calculated directly by
VASP, when compared to experimental studies. Percentage errors for other DFT studies are
shown for comparison.
Page 20
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 20
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S20
Supplementary Table 3: Calculated piezoelectric charge tensor components eij (in units of
C/m2), strain tensor components dik (pm/V), voltage tensor components gij (mV m/N), elastic
constants (GPa), and dielectric constants (unitless) of L-Asparagine.
Charge Tensor (C/m2)
(0 0 0 0.13 0 −0.09
0.11 0.05 0.09 0 −0.05 00 0 0 −0.01 0 0.04
)
Strain Tensor (pm/V)
(0 0 0 12.0 0 −13.02.1 2.0 1.1 0 −4.6 00 0 0 −0.9 0 5.7
)
Voltage Tensor (mV m /N)
(0 0 0 490 0 530
110 100 60 0 240 00 0 0 40 0 240
)
Elastic Stiffness Constant (GPa)
c11
52
c22
25
c33
83
c44
11
c55
11
c66
7
Dielectric Constant (unitless)
ε1 2.76
ε2 2.19
ε3 2.70
εr 2.55
Page 21
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 21
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S21
Supplementary Table 4: Calculated piezoelectric charge tensor components eij (in units of
C/m2), strain tensor components dik (pm/V), voltage tensor components gij (mV m/N), elastic
constants (GPa), and dielectric constants (unitless) of L-Aspartic acid.
Charge Tensor (C/m2)
(0 0 0 0.03 0 −0.05
−0.08 −0.05 0.06 0 0 00 0 0 −0.13 0 0.15
)
Strain Tensor (pm/V)
(0 0 0 3 0 3
−0.73 −1.85 −1.82 0 0 00 0 0 13 0 10
)
Voltage Tensor (mV m /N)
(0 0 0 126 0 12637 94 93 0 0 00 0 0 575 0 442
)
Dielectric Constant (unitless)
ε1 2.69
ε2 2.21
ε3 2.55
εr 2.48
Elastic Stiffness Constant (GPa)
c11
109
c22
27
c33
33
c44
10
c55
11
c66
15
Page 22
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 22
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S22
Supplementary Table 5: Calculated piezoelectric charge tensor components eij (in units of
C/m2), strain tensor components dik (pm/V), voltage tensor components gij (mV m/N), elastic
constants (GPa), and dielectric constants (unitless) of L-Methionine.
Charge Tensor (C/m2)
(0 0 0 −0.01 0 0.09
0.08 0.18 0.07 0 0.02 00 0 0 0.03 0 0
)
Strain Tensor (pm/V)
(0 0 0 5 0 13
2.42 6.9 4.1 0 5 00 0 0 15 0 0
)
Voltage Tensor (mV m /N)
(0 0 0 228 0 594
106 304 181 0 220 00 0 0 694 0 0
)
Dielectric Constant (unitless)
ε1 2.48
ε2 2.57
ε3 2.44
εr 2.50
Elastic Stiffness Constant (GPa)
c11
33
c22
26
c33
17
c44
2
c55
4
c66
7
Page 23
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 23
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S23
Supplementary Table 6: Calculated piezoelectric charge tensor components eij (in units of
C/m2), strain tensor components dik (pm/V), voltage tensor components gij (mV m/N), elastic
constants (GPa), and dielectric constants (unitless) of L-Cysteine.
Charge Tensor (C/m2)
(0 0 0 −0.03 0 0.09
0.15 0.36 −0.04 0 0.01 00 0 0 0 0 −0.01
)
Strain Tensor (pm/V)
(0 0 0 10.7 0 10.64.1 11.4 2.9 0 2.6 00 0 0 0 0 1.2
)
Voltage Tensor (mV m /N)
(0 0 0 487 0 487
175 487 124 0 111 00 0 0 0 0 55
)
Elastic Stiffness Constant (GPa)
c11
37
c22
32
c33
14
c44
3
c55
4
c66
9
Dielectric Constant (unitless)
ε1 2.55
ε2 2.64
ε3 2.49
εr 2.56
Page 24
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 24
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S24
Supplementary Table 7: Calculated piezoelectric charge tensor components eij (in units of
C/m2), strain tensor components dik (pm/V), voltage tensor components gij (mV m/N), elastic
constants (GPa), and dielectric constants (unitless) of L-Isoleucine.
Charge Tensor (C/m2)
(0 0 0 0.0049 0 −0.11
−0.05 0.24 −0.02 0 0.03 00 0 0 0.01 0 0.02
)
Strain Tensor (pm/V)
(0 0 0 12.25 0 18.32.6 10 1.1 0 10 00 0 0 25 0 3.33
)
Voltage Tensor (mV m /N)
(0 0 0 587 0 876
122 469 52 0 469 00 0 0 1233 0 164
)
Dielectric Constant (unitless)
ε1 2.36
ε2 2.41
ε3 2.29
εr 2.35
Elastic Stiffness Constant (GPa)
c11
19
c22
24
c33
19
c44
0.4
c55
3
c66
6
Page 25
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 25
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S25
Supplementary Table 8: Calculated piezoelectric charge tensor components eij (in units of
C/m2), strain tensor components dik (pm/V), voltage tensor components gij (mV m/N), elastic
constants (GPa), and dielectric constants (unitless) of L-Histidine.
Charge Tensor (C/m2)
(0 0 0 0.04 0 −0.09
0.04 0.02 −0.04 0 −0.02 00 0 0 0.01 0 −0.01
)
Strain Tensor (pm/V)
(0 0 0 4 0 181.6 1.25 0.43 0 6.67 00 0 0 2.5 0 2
)
Voltage Tensor (mV m /N)
(0 0 0 183 0 82681 63 21.8 0 340 00 0 0 103 0 80
)
Dielectric Constant (unitless)
ε1 2.46
ε2 2.23
ε3 2.75
εr 2.48
Elastic Stiffness Constant (GPa)
c11
25
c22
16
c33
92
c44
4
c55
3
c66
5
Page 26
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 26
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S26
Supplementary Table 9: Calculated piezoelectric charge tensor components eij (in units of
C/m2), strain tensor components dik (pm/V), voltage tensor components gij (mV m/N), elastic
constants (GPa), and dielectric constants (unitless) of L-Leucine.
Charge Tensor (C/m2)
(0 0 0 0 0 0.01
0.06 −0.18 0.03 0 0.05 00 0 0 −0.02 0 0.01
)
Strain Tensor (pm/V)
(0 0 0 0 0 12.52.7 8.6 1.5 0 12.5 00 0 0 20 0 12.5
)
Voltage Tensor (mV m/N)
(0 0 0 0 0 593
126 400 70 0 580 00 0 0 976 0 611
)
Dielectric Constant (unitless)
ε1 2.38
ε2 2.43
ε3 2.31
εr 2.37
Elastic Stiffness Constant (GPa)
c11
22
c22
21
c33
20
c44
1
c55
4
c66
8
Page 27
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 27
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S27
Supplementary Table 10: Piezoelectric charge tensor components eij (in units of C/m2),
strain tensor components dik (pm/V, equivalent to pC/N – we use the two units
interchangeably), and voltage tensor components gij (mV m/N) of γ-glycine. The format of
the tensor table is described further in main text Methods. For comparison, the shear
piezoelectric strain constants of other biomaterials10 include cellulose (0.15 pC/N), amylose
(2.0 pC/N), collagen (2.0 pC/N), and keratin (1.8 pC/N).
γ-glycine Charge Tensor (C/m2)
(0.04 -0.04 0 -0.07 0.10 0.030.03 -0.03 0 0.09 0.07 -0.04-0.01 -0.01 0.81 0 0 0
)
γ-glycine Strain Tensor (pm/V)
(1.6 -1.1 0 -5.8 7.2 5.61.2 -1.1 0 7.5 5.6 -5.6-0.8 -0.7 10.4 0 0 0
)
γ-glycine Voltage Tensor (mV m /N)
(73 51 0 -266 331 25755 51 0 344 257 25735 31 455 0 0 0
)
Page 28
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 28
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S28
Supplementary Table 11. Piezoelectric charge tensor components eij (in units of C/m2),
strain tensor components dik (pm/V), and voltage tensor components gij (mV m/N) of β-
glycine. These matrices emphasise how precise sample preparation and orientation can
maximise output voltage along shear planes for device applications.
β-glycine Charge Tensor (C/m2)
(0 0 0 -0.12 0 0.26
0.10 -0.13 0.13 0 -0.08 00 0 0 0.01 0 -0.05
)
β-glycine Strain Tensor (pm/V)
(0 0 0 15.8 0 195
1.8 -5.7 1.9 0 5.1 00 0 0 1.3 0 7.5
)
β-glycine Voltage Tensor (mV m /N)
(0 0 0 658 0 813194 -296 98 0 265 00 0 0 57 0 328
)
Page 29
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 29
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S29
Supplementary Table 12: Computed Packing Densities and Monoclinic Angles of the three
glycine polymorphs. Numbers of atoms include hydrogen atoms.
Amino Acid
Number of Atoms
(and Molecules) in
Unit Cell11-13
Crystal Unit
Cell Volume
(Å3)
Atomic (and
Molecular) Packing
Densities
Atoms (molecules)/Å3
Monoclinic
Angle (°)
α-glycine 40
(4) 321
0.125
(0.0125)
Not
applicable
β-Glycine 20
(2) 164
0.122
(0.0122) 112
γ-glycine 30
(3) 245
0.122
(0.0122)
Not
applicable
Page 30
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 30
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S30
Supplementary Table 13: Analysis of the unit cell parameters of beta and gamma glycine
crystals calculated with and without dispersion corrections. Values in parentheses are
literature experimental values, as referenced in Supplementary Figure 10.
No Correction DFT-D3 a (Å) 5.13 5.06
(5.08) b(Å) 6.39 6.20
(6.27) β- c(Å) 4.99 5.39
(5.38) glycine Cell Volume
(Å3) 164.03 155.2 (157)
E0 (eV) -119.147 -120.392 a (Å) 7.16 6.95
(7.04) γ- b(Å) 7.16 6.95
(7.04) glycine c(Å) 5.52 5.48
(5.48) Cell Volume
(Å3) 244.82 229.27 (235)
E0 (eV) -178.712 -180.546
Page 31
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 31
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S31
Supplementary Table 14: Elastic constant values of AlN, calculated with and without
dispersion corrections. All values are in GPa. Values in parentheses are literature
experimental values, as referenced in Supplementary Table 1.
No Correction DFT-D3 c11 377 430 (378)
c12 130 122 (137)
Aluminium c33 358 363 (392)
Nitride c44 112 113 (122)
c66 123 154 (121)
Page 32
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 32
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S32
Supplementary Table 15: Calculation of the average Young’s Moduli from the DFT data
for the three ambient polymorphs of glycine, using three methods of approximation. All three
methods give reasonable match with known experimental values14 for α and γ-glycine, so an
arithmetic mean is taken over the three analysis methods to estimate the previously unknown
Young’s Modulus of β-glycine.All units are in GPa.
Analysis Method Alpha Beta Gamma
Voigt-Reuss15,16 28.32 19.37 24.58
Orthorhombic Approximation17 (Nye) 31.21 13.84 31.51
Triclinic Approximation17 (Nye) 30.00 13.22 31.27
Experiment14 33 - 28
Predicted Average Young’s Modulus 29.84 15.48 27.78
Page 33
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 33
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S33
Supplementary Table 16: Comparison of piezoelectric charge coefficients calculated with
Finite Differences and Density Functional Perturbation Theory (DFPT) methods. All units
are C/m2. Deviations between Finite Differences and DFTP range from 0-15%, and the
maximum deviation is used as our theoretical error in this work.
e21 e22 e23 e33 e14 e16 e25 e34 e36
Beta Finite Differences 0.09 -0.11 0.11 - -0.12 0.25 -0.11 0.01 -0.06
Glycine DFPT 0.10 -0.13 0.13 - -0.12 0.26 -0.08 0.01 -0.05
Gamma Finite Differences 0.03 -0.03 - 0.83 -0.06 0.03 0.06 - -
Glycine DFPT 0.03 -0.03 - 0.81 -0.07 0.03 0.07 - -
Page 34
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 34
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S34
Supplementary Note: Computational benchmarking against known inorganic piezoelectric
crystals
The accuracy of the computational methods used in the simulation of polarisation, elastic
stiffness and piezoelectricity has been benchmarked with respect to three well-known
inorganic piezoelectric materials, namely aluminium nitride (AlN), zinc oxide (ZnO) and α-
quartz (SiO2). We have compared elastic and piezoelectric coefficients of these materials
computed by us with those obtained in previous experimental and computational studies3-9.
Table S1 shows the calculated piezoelectric and elastic constants of α-quartz, zinc oxide and
aluminium nitride. Table S2 shows the percentage error in our values, and those of other DFT
studies, when compared to the experimental works referenced in Table S1.
For aluminium nitride, our calculation generally predicts better piezoelectric and elastic
coefficients than those reported in the literature. There is one exception, the e15 coefficient,
which deviates from the experimental value by ~37%. This however compares well with the
deviation (35%) of the values predicted by current literature. For zinc oxide, the agreement of
our predictions and those predicted and measured in the literature is also good (Table S1 and
S2). This lends further confidence to our predicted values for the glycine biomolecules (Table
1).
We note that the results for both zinc oxide and aluminium nitride are both highly dependent
on the number of k points used in the calculations, with an 8x8x8 mesh giving a better match
with experimental results. Both crystals have a wurtzite structure, and belong to the space
group P 63/mc (number 186). This emphasises the value of more extensive k point sampling
(which requires more computational expense) when calculating second order derivatives of
hexagonal crystals, as has been observed in other computational studies 18-20.
Page 35
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 35
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S35
In the third and final set of benchmarking, we again find our predicted coefficients in α-
quartz (P3121) matching experimental values better than those predicted by previous DFT
studies, except for the coefficients c44 and c66, for which Ren et al8 predicts slightly better.
Yet, our computed values deviate from experimental values of c44 and c66 only by 3% and 9%
respectively. Interestingly, we note that both the elastic and piezoelectric constants for α-
quartz exhibit negligible dependence on the number of k points used. As both α-quartz and γ-
glycine form trigonal crystals, the accuracy obtained by us in predicting coefficients of α-
quartz build further confidence in the predicted physical constants for glycine polymorphs
(Table 1 main text).
Page 36
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 36
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S36
Supplementary Note: The effectiveness of density functional theory predictions of
piezoelectric glycine polymorphs
The excellent quantitative correspondence of DFT predicted stiffness, permittivity and
piezoelectricity tensors with those obtained from experiment conducted in the present study
and in the literature highlights the potential that DFT calculations now demonstrate to
identify materials with significant piezoelectric response, and to estimate the expected
magnitudes of individual piezoelectric constants. The recent upsurge in interest in biological
energy harvesting (and very promising results)21 highlights the need for efficient candidate
screening with DFT. This can pave the way for an exciting and more diverse range of
bioelectronic components.
We use DFT to quantify all eighteen components in the piezoelectric matrices of two non-
centrosymmetric polymorphs of glycine. Such quantitative predictions are desirable to guide
experiments and in the discussion of technological applications. These matrices can be
further separated into their ionic and electronic contributions. Induced dipole moments are
made up of contributions from both distortions of the symmetrical electronic charge
distributions relative to the nucleus of each ion (electronic), and from displacements of the
entire ion from its normal lattice site (ionic).
The DFT values are calculated at 0 K, so it should be noted that systems could exhibit a
deviation between predicted DFT values and properties measured at room temperature,
especially if there is a phase transition between 0 and 300 K. No such deviation has been
observed in this work, neither for the well-characterised inorganic crystals (in benchmarking
the periodic DFT methods used for the organic crystals), nor in the experimental confirmation
of the glycine longitudinal strain coefficients.
Page 37
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 37
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S37
Supplementary Note: Example calculation in the determination of piezoelectric constants
e16 and d16 of β-Glycine following methodologies outlined in supplementary references 22,23
for resonance piezoelectricity measurements in a monoclinic crystal using thickness mode
vibration measurements.
For all calculations:
ρ is defined as the average density of β-glycine, taken as 1610 kg/m3
Fa, is the anti-resonance frequency at the maximum impedance
Fr is the resonance frequency at minimum impedance
For capacitance measurements, crystal slice dimensions were
Thickness, t = 5.21 μm
Width, w = 212 μm
Length, l = 692 μm
Area, A = Area of the XY plane (w*l)= 1.467 x 10-7 m2
A. Calculation of Relative Permittivity, εr
ε11= C11tA
= C11tw x l
From direct measurement, capacitance in the 11 direction, C11 = 0.8 pF, so that
ε11=0.8 pF x 5.21μm212μm x 692μm
= 2.84 x 10-11 F/m
Therefore,
εr= ε11
ε0=
2.84 x 10-11
8.85 x 10-12
Page 38
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 38
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S38
Where ε0= permittivity of the free space = 8.85x 10-12 F/m and
εr= 3.21
B. Calculation of Elastic Stiffness Constant, c66
For measurements of the 16 coefficient, the thickness: width: length ratio should not be more
than 1:12:30 in accordance with CENELEC standards. The following dimensions were used
in cutting the samples from the needles
Thickness, t = 8.43 μm = 8.43x10-6 m
Width, w = 117 μm =117x10-6 m
Length = 265 μm = 265x10-6 m
Shear compliance s66 is given by:
s66= 1
4ρFa2l2
= 1
4 x 1610 x (1626000)2x (265×10−6)2
=1.20 GPa
c66=0.84 GPa
C. Calculation of Electromechanical Coupling Constant, k16
k162= π
2
FrFa
tan (π2
∆FFa
)
= π2
409
1626 tan (π
2 12171626
)
Page 39
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 39
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S39
= 1.57 x 0.25 x 2.40
= 0.94
k16 = √0.94 = 0.97
D. Calculation of Piezoelectric Charge Constant, e16
e16 = k16 √ε11c66
= 0.97 √2.84 x 10−11 x 0.84 x 109
= 0.15 C/m2
E. Calculation of Piezoelectric Strain Constant, d16
d16= e16
c66
= 0.15 C 𝑚𝑚−2
0.84 GPa
= 178 pm/V
Page 40
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 40
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S40
Supplementary Note: Exploiting piezoelectricity in β-glycine for technological applications
At first glance, it is natural to assume that such a hidden piezoelectric response, in a
metastable crystal, would be extremely difficult to utilise for technological applications.
However in recent years much progress has been made in the stabilisation of the
microcrystals, in the hope of exploiting the effective longitudinal and shear piezoelectric
constants in β-glycine. Values ranging from 6-13 pm/V were measured using Piezoresponse
Force Microscopy (PFM) 24, and are discussed in the main text. So far, these needles have
been successfully stabilised on (111) platinum thin films25, and also made into β-glycine
fibres via electrospinning26. Both of these processes, as discussed in the respective papers,
have huge technological potential, but do not exploit our measured value of 178 pm/V.
Hence the challenge for us was the technological exploitation of our previously unknown d16
and g16 values of 178 pm/V and 8131 V m/N respectively. For this, the crystal would have to
be electroded on its x-axis, which has an average length of 200 micron. The easiest way to
induce a shear stress along the z-axis would be to bend the crystal. For a single β-glycine
crystal, or even a poly-crystalline needle aggregate, the crystal would not be expected to
withstand any significant manual force. Supplementary Figure 15 shows our attempts to grow
-glycine layers on a flexible substrate,27 but we expect that extensive further research will be
required to fully realise exploitation of our predicted g16 value in this way.
It is worth noting that the γ-glycine device we report (to the best of our knowledge) is the first
energy harvesting device reported using undoped organic crystals, and is most certainly the
first energy harvesting device that uses undoped amino acid crystals. This proof of concept
device for a (relatively) modest predicted voltage constants (51 mV m/N), should stimulate
more research into the challenges surrounding organic energy harvesting, particularly for
Page 41
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 41
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S41
microneedle strucures such as β-glycine and L-isoleucine (predicted g values of 8131
and 1233 mV m/N respectively).
Page 42
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 42
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S42
Supplementary Note: Experimental precautions for characterisation of metastable β-glycine
As discussed above, there are many ways to stabilise β-glycine needles for technological
exploitation; however for our experimental verification the needles needed to be removed
from their growth substrate, sliced, and tested in ambient conditions. We found during our
work that coating any glass substrate in ethanol before coating with a 0.133 M solution of
glycine, yielded needles that were stable for 4-6 weeks. Outer regions transforming into α-
glycine or γ-glycine could be observed visually on a micron scale (main text Figure 3c), and
was confirmed at the molecular level using Raman Spectroscopy and X-Ray Diffraction.
Once individual needles were removed and sliced, they had to be placed on a slide and
positioned in the probe station. Crystals would be in ambient conditions for an average of five
minutes before microprobe alignment, which could take another ten minutes. Extraction of
correctly compensated resonance data using LabVIEW takes approximately three minutes.
Only Raman Spectroscopy, calibrated and ready for analysis, could be used to confirm the
presence of β-glycine post-measurement. The logistics for X-Ray Diffraction characterisation
require more time than β-glycine has been known to remain stable for in ambient conditions.
X-Ray Diffraction was used only to confirm the presence of pre-measurement β-glycine
which had been grown on glass substrates and stored below room temperature and humidity.
The fast transformation reported by Isakov et al. (reference 46 in the main text) was induced
by rapidly increasing the relative humidity, and the crystals were grown on a different
substrate. We have not observed such a rapid transformation during our work. When growing
crystals by droplet evaporation, the edges of the droplet do transform into α-glycine crystals
in a matter of hours (see Figure 3c) but the macroscopic needles at the centre of the droplet,
and those grown in petri dishes with ethanol, have remained stable for much longer than is
reported in the literature. When grown on platinum25 substrates β-glycine crystals have been
stabilised for one month. For our work we have microcrystals grown in petri dishes that have
Page 43
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 43
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S43
remained stable after three months. In terms of crystals that are suitable for resonance
measurements (neither too thin nor too strongly adhered to the substrate) we see no signs of
transformation for 2-3 weeks when stored slightly below room temperature and humidity.
However, given that the crystals do spend finite periods of time in environments with a
higher relative humidity than that in which they are stored, it is definitely possible that
surface transformations could take place on the crystal slices. A complete transformation to
the γ or α form can be observed using optical microscopy or Raman spectroscopy, as shown
in Supplementary Figure 3. This can happen to sporadically to needles that are removed from
the substrate, or do not adhere strongly to the substrate during crystallisation. We feel it is
more realistic to acknowledge the potential development of stable polymorphs within β-
needles during measurement, which can lead to a reduction in our measured piezoelectric
constant (d16 = 0 pm/V and 5.6 pm/V) for α and γ glycine respectively).
Page 44
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 44
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S44
Supplementary Note: Grimme DFT-D3 Dispersion Corrections
For β and γ glycine, convergence of unit cell volumes (and individual lattice parameters) with
respect to cut-off radius is achieved at 12Å and 8Å respectively. At this convergence point
we find a reduction in energy of ~1.6 eV and a corresponding reduction of ~4% in the cell
volume (Supplementary Figure 10) to just below the experimental volume (Supplementary
Table 15) for both polymorphs. Van der Waals binding between the molecules results in a
proportional increase in elastic tensor values (Supplementary Figure 11) and a corresponding
decrease in piezoelectric constants (discussed in the main text), which moves the predicted
values away from the measured experimental values (main text Fig. 2).
As a control, we tested also the sensitivity of AlN dispersion corrections to cut-off values,
AlN28 being a far less malleable, inorganic structure that requires at least 14 GPa29 to undergo
a pressure induced transformation. The addition of vdW forces reduces the cell volume by
just 1%, but the calculated elastic constants change significantly. We find a percentage
deviation from experimental values of 14%, 7% and 27% for c11, c33 and c66 respectively,
larger than the corresponding deviations in the purely electronic DFT calculations of 0%, 9%,
and 2% (see Supplementary Figure 11 and Supplementary Table 1).
We suspect that this discrepancy between experimental measurements and predictions could
stem from increased external pressures acting on the crystals in the dispersion-corrected
calculations, which could possibly be used to simulate transformations between glycine
polymorphs in future studies.
Page 45
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 45
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S45
Supplementary References
1 Weissbuch, I., Torbeev, V. Y., Leiserowitz, L. & Lahav, M. Solvent effect on crystal polymorphism: why addition of methanol or ethanol to aqueous solutions induces the precipitation of the least stable β form of glycine. Angewandte Chemie International Edition 44, 3226-3229 (2005).
2 Ferrari, E. S., Davey, R. J., Cross, W. I., Gillon, A. L. & Towler, C. S. Crystallization in polymorphic systems: the solution-mediated transformation of β to α glycine. Crystal growth & design 3, 53-60 (2003).
3 Supryadkina, I., Abgaryan, K., Bazhanov, D. & Mutigullin, I. Ab initio study of macroscopic polarization of AlN, GaN and AlGaN. physica status solidi (c) 11, 307-311 (2014).
4 Zoroddu, A., Bernardini, F., Ruggerone, P. & Fiorentini, V. First-principles prediction of structure, energetics, formation enthalpy, elastic constants, polarization, and piezoelectric constants of AlN, GaN, and InN: Comparison of local and gradient-corrected density-functional theory. Physical Review B 64, 045208 (2001).
5 Tsubouchi, K., Sugai, K. & Mikoshiba, N. in 1981 Ultrasonics Symposium. 375-380 (IEEE).
6 Chowdhury, R., Adhikari, S. & Scarpa, F. Elasticity and piezoelectricity of zinc oxide nanostructure. Physica E: Low-Dimensional Systems and Nanostructures 42, 2036-2040 (2010).
7 Kobiakov, I. Elastic, piezoelectric and dielectric properties of ZnO and CdS single crystals in a wide range of temperatures. Solid State Communications 35, 305-310 (1980).
8 Ren, J. et al. Structure, elastic and piezoelectric properties of A 3 BO 7 (A= Ga, Al; B= P, As) compounds: A DFT study. Computational Materials Science 106, 1-4 (2015).
9 Bechmann, R. Elastic and piezoelectric constants of alpha-quartz. Physical review 110, 1060 (1958).
10 Fukada, E. History and recent progress in piezoelectric polymers. Ultrasonics, Ferroelectrics, and Frequency Control, IEEE Transactions on 47, 1277-1290 (2000).
11 Iitaka, Y. Crystal Structure of [beta]-Glycine. Nature 183, 390-391 (1959).
12 Albrecht, G. & Corey, R. B. The crystal structure of glycine. Journal of the American Chemical Society 61, 1087-1103 (1939).
13 Iitaka, Y. The crystal structure of γ-glycine. Acta Crystallographica 11, 225-226 (1958).
14 Azuri, I. et al. Unusually Large Young’s Moduli of Amino Acid Molecular Crystals. Angewandte Chemie International Edition 54, 13566-13570 (2015).
15 Voigt, Woldemar. Lehrbuch der kristallphysik (mit ausschluss der kristalloptik). Springer-Verlag, 2014.
Page 46
© 2017 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
NATURE MATERIALS | www.nature.com/naturematerials 46
SUPPLEMENTARY INFORMATIONDOI: 10.1038/NMAT5045
S46
16 Reuss, A. Calculation of the flow limits of mixed crystals on the basis of the plasticity of monocrystals. Z. Angew. Math. Mech 9, 49-58 (1929).
17 Nye, J. F. Physical properties of crystals: their representation by tensors and matrices. (Oxford university press, 1985).
18 Ding, Y., Chen, M. & Xiao, B. Anisotropy in lattice thermal conductivity tensor of bulk hexagonal-MT 2 (M= W, Mo and T= S and Se) by first principles phonon calculations. RSC Advances 6, 7817-7828 (2016).
19 Sun, Z., Li, S., Ahuja, R. & Schneider, J. M. Calculated elastic properties of M 2 AlC (M= Ti, V, Cr, Nb and Ta). Solid state communications 129, 589-592 (2004).
20 Yu, J. & Guo, W. Two-Dimensional Hexagonal Beryllium Sulfide Crystal. The journal of physical chemistry letters 4, 1856-1860 (2013).
21 Nguyen, V., Zhu, R., Jenkins, K. & Yang, R. Self-assembly of diphenylalanine peptide with controlled polarization for power generation. Nature Communications 7, 13566 (2016).
22 Chen, Q. & Wang, Q.-M. The effective electromechanical coupling coefficient of piezoelectric thin-film resonators. Applied Physics Letters 86, 022904 (2005).
23 Bechmann, R. Determination of the elastic and piezoelectric coefficients of monoclinic crystals, with particular reference to ethylene diamine tartrate. Proceedings of the Physical Society. Section B 63, 577 (1950).
24 Seyedhosseini, E. et al. Tip-induced domain structures and polarization switching in ferroelectric amino acid glycine. Journal of Applied Physics 118, 072008 (2015).
25 Seyedhosseini, E. et al. Growth and Nonlinear Optical Properties of β-Glycine Crystals Grown on Pt Substrates. Crystal Growth & Design 14, 2831-2837 (2014).
26 Isakov, D. et al. Production of polar β-glycine nanofibers with enhanced nonlinear optical and piezoelectric properties. Crystal Growth & Design 11, 4288-4291 (2011).
27 Hu, Y., Zhang, Y., Xu, C., Zhu, G. & Wang, Z. L. High-output nanogenerator by rational unipolar assembly of conical nanowires and its application for driving a small liquid crystal display. Nano letters 10, 5025-5031 (2010).
28. Ott, H, Das Gitter des Aluminiumnitrids (Al N), Zeitschrift fuer Physik, 1924, 22, 201-214
29. Xia, Qing, Hui Xia, and Arthur L. Ruoff. "Pressure‐induced rocksalt phase of aluminum nitride: A metastable structure at ambient condition." Journal of applied physics 73.12 (1993): 8198-8200.