Sunlight Active g-C3N4-based Mn+ (M = Cu, Ni, Zn, Mn) – Promoted Catalysts:

Sharing of Nitrogen Atoms as a Door for Optimizing Photo-activity

Mario J. Muñoz-Batista,1,2 Leandro Andrini,3 Félix G. Requejo,3,4

Maria N. Gómez-Cerezo,1 Marcos Fernández-García, 1,* Anna Kubacka,1,*

1 Instituto de Catálisis y Petroleoquímica, CSIC, C/Marie Curie 2, 28049-Madrid, Spain2 Department of Chemical Engineering, Faculty of Sciences, University of Granada.

Avda. Fuentenueva, s/n 18071, Granada, Spain.3 Instituto de Fisicoquímica Teórica y Aplicada (INIFTA), Dpto. de Química, Fac. de Ciencias

Exactas, (UNLP-CONICET), 64 y Diagonal 113 (1900), La Plata, Argentina.4 Dpto. de Física, Fac. de Ciencias Exactas, Universidad Nacional de La Plata, UNLP,

49 y 115 (1900), La Plata, Argentina.

1

Catalytic set-up and details

Figure S1. Upper, Left: Photocatalytic annular reactor scheme. (1) gas inlet, (2) gas outlet, (3) lamps, (4) catalyst (brown) sample. q¿ radiation flow on the surface of the

sample (red), qn radiation flow from the lamps (blue). Upper, Right: Center of coordinates located at the sample (defined by coordinates xs, ys, zs). Down, Coordinate system to define the integration limits of radiation Model. (Left) φmin and φmax. (Right)

Ѳmin and Ѳmax.

Table S1. Comparison of catalytic observables for the Zn/g-C3N4 sample carried out under (our) standard conditions (24 h) and for a prolonged test (72 h).

Test Quantum Efficiency / % CO2 Selectivity / %Standard 2.01 ± 0.14 x 10-10 (UV) 38 ± 2.8(UV)

1.59 ± 0.12 x 10-10 (ST) 39± 2.9 (ST)Long Term 2.05 ± 0.16 x 10-10 (UV) 40 ± 2.9(UV)

1.49 ± 0.13 x 10-10 (ST) 38± 2.7 (ST)

2

Integration limits of equation 2 are summarized in equations S1-S7 and can be graphically visualized in Figure S1.

φ1=tan−1( XL−XsY L−Ys ) (S1)

φ2=sin−1( RL

( X L−X s )2+ (Y L−Y s )2 ) (S2)

φmin=φ1−φ2 (S3)

φmax=φ1+φ2 (S4)

Ѳmin(φ)=cos−1 −Zs

( XLm ( φ )−X s )2+(Y Lm (φ )−Y s )2+Z s2 (S5)

Ѳmax (φ)=cos−1 ZL−Z s

( XLm ( φ )−X s )2+(Y Lm (φ )−Y s )2+Z s2 (S6)

Where:

X Lm (φ )=X L+( X s−Y L) cos φ2+(Y L−Y s )¿

(S7)

Y Lm (φ )=Ysi+(Y L−Y s ) cosφ2+( X s−XL )¿

(S8)

Where symbols were previously defined (main text) except X L, Y L, ZL which are the coordinates of the points located on the surface of the lamp.

3

Characterization results

Zn Cu

Figure S2. Electron diffraction analysis of selected M/g-C3N4 samples.

292 290 288 286 284 282 280406 404 402 400 398 396 394

g-C3N4

A Mn+/g-C3N4C 1s

B.E. / eV

B

g-C3N4

Zn

Ni

Cu

N 1s

B.E. / eV

Mn

Zn

Ni

Cu

Mn

Figure S3. C 1s (A) and N 1s (B) XPS peaks for M/g-C3N4 samples and g-C3N4

reference.

4

Zn

A

9650 9660 9670 9680

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Zn-gC3N

4

Nor

m. a

bs. (

arb.

uni

t.)

Energy (eV)

B

Figure S4. (A) Model of the Zn organometallic complex used (Zn atoms indicated with arrow, blue-gray color; N atoms, blue color; O atoms, red color; C atoms, black color; H atoms, gray color). (B) XANES simulation of the model compound using a full multiple scattering with muffin-tin potentials (dotted line); and full atomic potential obtained via

Schrödinger equation (dotted-dashed line).

5

EXAFS signal analysis

To isolate the EXAFS signal we follow standard procedures (1). The pre-edge background was discounted using a realistic decay of energy in absorption processes defined by the Victoreen function, a E−3+b. The EXAFS-signal, χ, was subsequently normalized by μ0(E) – μb(E), where the post-edge background μ0 was considered as a smoothing spline (the μ0 curve is constructed from the experimental μ spectrum by smoothing out the EXAFS wiggles), using an equidistant k mesh from kmin = 1 and kmax = 13 Å-1.

The analysis of the EXAFS signal was carried out using a standard formulation, equation S1.

χ (k )=S02∑

i

N i f i (k )k Ri

2 sin [2 k R j+δ j (k )]exp (−2 k2 σ i2) (S1)

Where χ (k ) is the simple scattering EXAFS signal, S02 is the many-body factor, f (k ) and

δ (k ) are the amplitude and phase functions, respectively, N is the number of neighboring atoms, R is the distance to the neighboring atom, and σ 2 is the disorder in the neighbor distance shell. Though somehow complicated, the EXAFS equation allows us to determine N, R, and σ 2 using a minimization procedures and knowing the scattering amplitude f(k) and phase-shift δ(k). Furthermore, since scattering factors depend on the atomic number (Z) of the neighboring atom, EXAFS is also sensitive to the atomic species around the absorbing atom (1).

To fit the EXAFS using equation S1 we consider:1. There are not multiple scattering contributions in all cases reported. This

assumption is inherent to equation 1 and is commonly used (1).2. The bulk ZnO (FEFF) model is as simple as possible, i.e.: feff0001.dat for 1 Zn-

O at 1.7964 Å, feff0002.dat for 3 Zn-O at 2.0423 Å, feff0003.dat for 6 Zn-Zn at 3.2090 Å, and feff0004.dat for 6 Zn-Zn at 3.2495 Å;

3. For the Zn/g-C3N4 sample; using the Nyquist theorem (1) we calculate

N ind=2 Δk Δ R

π+2. N ind is the number of independent points; ca. 19 in our case;

¿ ν=N ind−P is the degree of freedom; 7 in our case, with P is the number of fitting parameters. This fully justifies the possibility of using a four shell model for fitting.

4. Concerning fitting parameters, we note that E0 positioning is global and is common for all the coordination shells. However, the calculated scattering amplitudes and phases may have shifts in Fermi energy (due to limitations in the theory to model electronic details of the materials). Moreover, the shifts may be different for different types of atoms although this difference should not be big. It is typically about 1 eV with self-consistent calculations and about 3 eV with

6

overlapped atom potentials (1,2). Of course, the energy shifts for the same atoms in different coordination shells should be constrained as equal.The intrinsic loss measured by S0

2 was obtained averaging the results of two methods by: i) adjusting the experimental signal of the Zn-metallic (0.949), and ii) using FEFF (0.952).

5. The k3-weighted EXAFS-signal were Fourier transformed from the k space to

the rspace to give a radial function ϕ (r )= 1√2 π ∫

kmin

kmax

k3 χ (k )W ( k ) ei 2 kr dk with a

Hanning window W (k ) as a function applied to the first and last 5% of the k-space data to smooth the finite transform between the lower, k min, and upper, k max

, k values of the χ (k) (3).

Experimental errors in EXAFS curve. To determine the error in the EXAFS curve, the ε k module was calculated with the VIPER program (2). There are two major contributions to the experimental errors in χ (k): those due to (i) the (statistical, gaussian one) of the measurement(s) and (ii) the (non statistical) uncertainty in calculating μ0. The use of the two sources of error is normally dismissed in the literature but it is here consider to fully account for the error in the EXAFS results. The errors in the oscillating signal are due to uncontrolled (statistical and not statistical) noise, affecting the Fourier Transform (FT) of the signal (equation S2).

ε=|χ (k ) . BFT (k)kw | (S2)

Where k w is the weighted of the signal (w=3, in our case), and BFT (k ) is the back Fourier Transform signal, here utilized with a k min cutoff of 0.90 (rmax=¿6 Å).

For the experimental Zn K-edge EXAFS curve of the Zn/g-C3N4 sample, the following results are obtained:

ε min(k) = 7.115x10-7

ε max(k) = 2.186x10-3

Simple mean of ε (k ) = 5.134x10-4

Quadratic mean of ε (k ) = 7.256x10-4

Inverse quadratic mean of ε (k ) = 1.211x10-5

Using the ε (k ) values and appropriate algebra (from equation S1), the error in the EXAFS parameters are obtained as detailed in ref. (2).

7

Table S2. Microstrain of the reference (g-C3N4) and catalysts.

Sample Micro-strain (×10−3)g-C3N4 2.1 ± 0.2

Ni/g-C3N4 2.0 ± 0.2Cu/g-C3N4 2.0 ± 0.2Zn/g-C3N4 2.1 ± 0.2Mn/g-C3N4 2.1 ± 0.2

200 300 400 500 600 700 800 900

Kub

elka

-Mun

k

Wavelength / nm

g-C3N

4

Cu/g-C3N

4

Mn/g-C3N

4

Ni/g-C3N

4

Zn/g-C3N

4

Figure S5. UV-visible spectra of the studied samples and references.

A B C D E

F G H I J

Figure S6. Superficial rate of photon absorption (Einstein cm-2 s-1) for UV (upper row) and Sunlight-type (lower row) illumination. (A,F) g-C3N4; (B,G); Ni/g-C3N4;(C,H)

Cu/g-C3N4;(D,I) Mn/g-C3N4;(E,J) .

8

0 1 2 3 4 50

250

500

750

1000

1250

1500

1750

2000

. DM

PO

-O- 2 a

dduc

t int

esity

Time / min.

B

Figure S7. Evolution of the radical species signal intensity for DMPO-OH and DMPO-O2 adducts as a function of irradiation time under UV illumination. Lines correspond to the rate of hydroxyl (A) or superoxide (B) radical formation for all samples under study.

Table S3. Values of initial rates of radical formation (rOH •and rO2•; spin min-1) for the samples as measured with EPR.a

Sample OH • (UV) OH • (ST)b O2•−¿ ¿ (UV) O2

•−¿ ¿ (ST)b

g-C3N4 905 200 1570 190Ni/g-C3N4 760 180 2040 230Cu/g-C3N4 820 195 1480 210Zn/g-C3N4 3210 320 910 180Mn/g-C3N4 1270 280 600 170

a Values are multiplied by a 1.5 x 10-4 constant; standard error: 15.5 %.

b ST: Sunlight-type illumination.

9

0 2 4 6

0

500

1000

1500

2000

2500

Mn/g-C3N4

Zn/g-C3N4

Cu/g-C3N4

Ni/g-C3N4

g-C3N4

. DMPO

-OH

addu

ct in

tesit

y

Time / min.

A

References

10

1 M. Newville. Fundamentals of XAFS. Reviews in Mineralogy and Geochemistry, 78(1) (2014) 33-74.2 See VIPER manual (https://www.albasynchrotron.es/en/beamlines/bl22-claess/software).3 Stern, E. A., Sayers, D. E., & Lytle, F. W. (1975). Extended x-ray-absorption fine-structure technique. III. Determination of physical parameters. Physical Review B, 11(12), 4836