Supplementary material Detection of nonlinearities in electrochemical impedance spectra by Kramers Kronig Transforms Authors: Fathima Fasmina, Ramanathan Srinivasana* Address: a Department of Chemical Engineering, Indian Institute of Technology-Madras, Chennai 600036, India. Phone : +91 44 2257 4171 Fax: +91 44 2257 0509 Email: srinivar@iitm.ac.in * corresponding author Index: 1. Fig. S1. (a) Sum of squares vs. number of Voigt elements for the data in Fig. 7. Normalized residual error resulting from a measurement model fit with 95% confidence intervals (b) Vac0 = 300 mV, real and (c) Vac0 = 300 mV, imaginary parts. 2. Fig. S2. (a) Log |iF| as a function of dc potential for the data in Fig. 8 (b) sum of squares vs. number of Voigt elements. Normalized residual error resulting from a measurement model fit with 95% confidence intervals (c) Vac0 = 300 mV, real and (d) Vac0 = 300 mV, imaginary parts. (e) Residual error resulting from linear KKT fit at Vac0 = 300 mV. 3. Fig. S3. Impedance spectrum for three step reaction with k10 = 10-10 mol cm-2 s-1, b1 =15 V-1, k20 = 10-12 mol cm-2 s-1, b2 = 30 V-1, k30 = 2 ×10-8 mol cm-2 s-1, b3 = 0 V-1, Γ = 10-8 mol cm-2 and Vdc = 0.25 V. (a) Complex plane plots of impedance spectra Vac0 = 1 mV and 150 mV (b) log|i F| as a function of dc potential. Bode plots of (c) |Z| and (d) at Vac0 = 150 mV with results of direct integration of KKT. (e) sum of squares vs. number of Voigt elements. Normalized residual error resulting from a measurement model fit with 95% confidence intervals (f) Vac0 = 150 mV, real and (g) Vac0 = 150 mV, imaginary part. (h) Residual error resulting from linear KKT fit at Vac0 = 300 mV. 4. Fig. S4. Example illustrating that when log|iF| vs. dc potential is linear, direct integration KKT and measurement model analysis do not flag the nonlinearity, but linear KKT successfully flags the nonlinearity. Impedance spectrum for two step reaction with k10 = 10-12 mol cm-2s-1, b1 =10 V, k20 = 10-9mol cm-2s-1, b2 = 9 V-1, and Γ = 10-8mol cm-2. (a) Complex plane plots of impedance 1 spectra Vac0 = 1 mV and 250 mV with Vdc = 0.7 V (b) log |iF| as a function of dc potential. Bode plots of (c) |Z| and (d) at Vac0 = 250 mV with results of direct integration of KKT (e) Residual error resulting from measurement model fit at Vac0 = 250 mV. The residuals were normalized with the magnitude of impedance (f) Residual error resulting from linear KKT fit at Vac0 = 250 mV. 5. Fig. S5 Example illustrating that when log|iF| vs. dc potential is nonlinear, all three data validation methods successfully flag the nonlinearity. Impedance spectrum for two step reaction with k10 = 10-9 mol cm-2s-1, b1 =3 V-1, k20 = 10-11 mol cm-2s-1, b2 = 15 V-1, and Γ = 10-8 mol cm-2. (a) Complex plane plots of impedance spectra Vac0 = 1 mV and 300 mV with Vdc = 0.5 V (b) log |iF| as a function of Vdc. Bode plots of (c) |Z| and (d) at Vac0 = 300 mV with results of direct integration of KKT (e) Sum of squares vs. number of Voigt elements. Normalized residual error resulting from a measurement model fit with 95% confidence intervals (f) Vac0 = 300 mV, real and (g) Vac0 = 300 mV, imaginary parts. (h) Residual error resulting from linear KKT fit at Vac0 = 300 mV. 6. Fig. S6. Data validation for the spectra given in Fig 9b after adding an equivalent resistance of 100 -cm2 in series. Bode plots of (a) the magnitude of admittance |Y| and (b) at Vac0 = 1 mV, with results of direct integration of KKT using data in the admittance form. 7. Fig. S7. Data validation for the spectra given in Fig. 9b. (a) Residual errors in measurement model fit at Vac0 = 300 mV. The residuals were normalized with the magnitude of impedance (b) Residual errors resulting from linear KKT fit performed in admittance mode, at Vac0 = 300 mV 8. Table S1. Summary of the results Fig. S1. 6 10 Sum of squares Vac0 = 1 mV Vac0 = 300 mV 4 10 2 10 0 10 -2 10 1 2 3 4 Number of Voigt elements Fig.S1 (a) 5 Vac0 = 300 mV (ZIm - ZMM ) / ZIm Im (ZRe - ZMM ) / ZRe Re 0.2 2 Voigt elements 0 -0.2 -3 10 10 0 frequency/ Hz Fig.S1 (b) 10 3 Vac0 = 300 mV 2 Voigt elements 0 -5 -3 10 10 0 10 3 frequency/ Hz Fig.S1 (c) Fig. S2. 10 10 -2 log ( iF / A cm ) Sum of squares -6 5 10 -9 Vac0 = 1 mV Vac0 = 300 mV 0 10 -12 -5 10 0.1 0.5 V /V 0.9 1 2 3 4 5 Number of Voigt elements dc Fig.S2 (a) Fig.S2 (b) 1.5 / ZIm Vac0 = 300 mV 3 Voigt elements ZMM ) Im 0 0 (ZIm - (ZRe - ZMM ) / ZRe Re 0.5 Vac0 = 300 mV 3 Voigt elements -0.5 -3 10 10 0 10 3 frequency/ Hz real, imag/ % Vac0 = 300 mV 0 real part imaginary part 0 Fig.S2 (e) 0 10 Fig.S2 (d) 20 10 frequency/ Hz 10 frequency/ Hz Fig.S2 (c) -75 -3 10 -1.5 -3 10 3 10 3 Fig. S3. 230 -6 log ( iF / A cm ) Vac0 = 1mV -ZIm / cm 2 -2 Vac0 = 150mV -9 -12 0 -20 0 Z Re / cm 0.1 230 2 0.5 V /V 0.9 dc Fig.S3 (a) Fig.S3 (b) 3 10 130 simulated data KKT by integration |Z| / cm 2 simulated data KKT by integration Vac0 = 150 mV 0 10 -3 10 0 0 -3 10 3 10 10 frequency / Hz 6 Sum of squares Vac0 = 1 mV Vac0 = 150 mV 2 10 0 10 -2 10 1 2 3 10 Fig.S3 (d) 10 4 0 10 frequency / Hz Fig.S3 (c) 10 Vac0 = 150 mV 3 4 Number of Voigt elements Fig.S3 (e) 5 5 Vac0 = 150 mV (ZIm - ZMM ) / ZIm Im (ZRe - ZMM ) / ZRe Re 0.05 3 Voigt elements 0 -0.05 -3 10 10 0 10 3 frequency/ Hz real, imag/ % real part imaginary part 0 Vac0 = 150 mV 0 Fig.S3 (h) 0 -5 -3 10 10 0 10 Fig.S3 (g) 10 10 frequency/ Hz 3 Voigt elements frequency/ Hz Fig.S3 (f) -5 -3 10 Vac0 = 150 mV 3 10 3 Fig. S4. 600 -6 -2 log ( iF / A cm ) Vac0 = 1 mV -ZIm / cm 2 Vac0 = 250 mV -10 -14 0 0 Z Re / cm 0.1 600 2 0.5 V /V 0.9 dc Fig.S4 (a) Fig.S4 (b) 130 3 simulated data KKT by integration simulated data KKT by integration |Z| / cm 2 10 Vac0 = 250 mV 0 10 -3 10 0 0 -3 10 3 10 Vac0 = 250 mV 10 frequency / Hz 0 3 10 10 frequency / Hz Fig.S4 (c) Fig.S4 (d) 1 real part imaginary part 0 Vac0 = 250 mV real, imag/ % real, imag/ % 1 real part imaginary part 0 Vac0 = 250 mV 2 Voigt elements -1 -3 10 0 10 frequency/ Hz Fig.S4 (e) 3 10 -1 -3 10 0 10 frequency/ Hz Fig.S4 (f) 3 10 Fig S5. -5 300 -2 log ( iF / A cm ) Vac0 = 1mV -ZIm / cm 2 Vac0 = 300mV -8 -11 0 0 Z Re / cm 2 0.1 300 0.5 V /V 0.9 dc Fig.S5 (a) Fig.S5 (b) 4 10 130 simulated data KKT by integration |Z| / cm 2 simulated data KKT by integration Vac0 = 300 mV 0 10 -3 10 0 0 -3 10 3 10 10 frequency / Hz Fig.S5 (d) 6 Sum of squares 10 4 10 2 Vac0 = 1 mV Vac0 = 300 mV 0 10 -2 10 1 2 0 10 frequency / Hz Fig.S5 (c) 10 Vac0 = 300 mV 3 Number of Voigt elements Fig.S5 (e) 4 3 10 10 Vac0 = 300 mV (ZIm - ZMM ) / ZIm Im (ZRe - ZMM ) / ZRe Re 0.08 2 Voigt elements 0 -0.08 -3 10 10 0 10 3 frequency/ Hz real, imag/ % Vac0 = 300 mV real part imaginary part 0 10 frequency/ Hz Fig.S5 (h) 0 -10 -3 10 10 0 10 Fig.S5 (g) 0 -20 -3 10 2 Voigt elements frequency/ Hz Fig.S5 (f) 5 Vac0 = 300 mV 3 10 3 Fig. S6. V ac0 -1 V = 1 mV ac0 200 100 -2 10 |Y| / -1 cm-2 10 = 1 mV Simulated data +100 KKT by Integration -3 10 -3 10 0 10 Frequency / Hz Fig.S6 (a) 3 10 0 -100 Simulated data +100 KKT by Integration -3 10 0 10 Frequency / Hz Fig.S6 (b) 3 10 Fig. S7. real, imag/ % 35 real part imaginary part Vac0 = 300 mV 0 3 Voigt elements -35 -3 10 0 10 frequency/ Hz Fig.S7 (a) 3 10 Fig.S7 (b) Table S1. Parameter sets Is log(iF) vs. Vdc nonlinear? Does KKT detect nonlinearity? Does linear KKT detect nonlinearity? True Does measurement model detect nonlinearity? True Fig. S3 True Fig. S4 Fig. S5 Estimated True 80.8 False False True True True True True True 240.0 225.5 Rct,NL (cm)