www.sciencemag.org/cgi/content/full/310/5751/1166/DC1

Supporting Online Material for

Porous, Crystalline, Covalent Organic Frameworks

Adrien P. Côté, Annabelle I. Benin, Nathan W. Ockwig, Michael O’Keeffe, Adam J. Matzger, Omar M. Yaghi*

*To whom correspondence should be addressed. E-mail: oyaghi@umich.edu

Published 18 November 2005, Science 310, 1166 (2005)

DOI: 10.1126/science.1120411

This PDF file includes:

Materials and Methods Figs. S1 to S37 Tables S1 to S4

Materials and Methods for

Porous, Crystalline Covalent Organic Frameworks

Adrien P. Côté†, Annabelle I. Benin†, Nathan W. Ockwig†, Michael O’Keeffe‡, Adam J. Matzger†, Omar M. Yaghi†,*

Materials Design and Discovery Group

†Department of Chemistry University of Michigan

930 North University Avenue Ann Arbor, Michigan, 48109-1055, USA

Tel: 734-615-2146, Fax: 734-615-9751

*Email: oyaghi@umich.edu

‡ Department of Chemistry and Biochemistry Arizona State University

Tempe, Arizona, 85287-1604, USA

Materials and Methods Table of Contents Section S1 Full synthetic procedures for the preparation of COF-1 and COF-5

and activation methods for gas adsorption measurements S2

Section S2 FT-IR Spectroscopy of Starting Materials, Model Compounds, and

COFs S5

Section S3 Scanning Electron Microscopy Imaging (SEM) and Energy

Dispersive X-ray (EDX) analysis of COF-1 and COF-5 S21

Section S4 11B MAS and 13C CP-MAS Nuclear Magnetic Resonance Studies for

COF-1 and COF-5 S23

Section S5 Structural Models and X-ray Analyses S26 Section S6 Low Pressure (0 – 1.0 bar) Gas Adsorption Measurements for

COF-1 and COF-5 at 77, 87, and 293 K S33

Section S7 Thermalgravimetry S46 Section S8 Mass spectrum of guests extracted from COF-5 prior to gas

adsorption analysis S48

S1

Materials and Methods Section S1: Full synthetic procedures for the preparation of COF-1 and COF-5 and activation methods for gas adsorption measurements. General Synthetic Procedures: All starting materials and solvents, unless otherwise

noted, were obtained from the Aldrich Chemical Co. and used with out further

purification. Transfer of all reagents was performed in an ambient laboratory air

atmosphere with no precautions taken to exclude oxygen or atmospheric moisture. Pyrex

glass tubes charged with reagents and flash frozen with LN2 were evacuated using a

Schlenk line by fitting the open end of the tube inside a short length of standard butyl

rubber hose that was further affixed to a ground glass tap which could be closed to isolate

this assembly from the dynamic vacuum when the desired internal pressure was reached.

Tubes were then sealed under this static vacuum using an oxygen-propane torch. Sealing

the tubes at 150(5) mtorr leads to optimal yields and crystallinity for both COFs, outside

this pressure range yields diminished slightly at lower pressures and notably at higher

pressures. We rationalize this observation on the fraction of H2O that becomes volatilized

into the headspace of the tube thereby shifting the equilibrium of the reaction either

towards amorphous products (lower pressures) and starting material (higher pressures).

Synthesis of COF-1. A Pyrex tube measuring o.d. × i.d. = 10 × 8 mm2 was charged with

1,4-benzene diboronic acid (BDBA) (25 mg, 0.15 mmol, Aldrich) and 1 mL of a 1:1 v:v

solution of mesitylene:dioxane. The tube was flash frozen at 77 K (LN2 bath), evacuated

to an internal pressure of 150 mtorr and flame sealed. Upon sealing the length of the tube

was reduced to 18 cm. The reaction mixture was heated at 120 ºC for 72 h yielding a

white solid at bottom of the tube which was isolated by filtration and washed with

acetone (30 mL). Yield: 17 mg, 71 % for (C3H2BO)6•(C9H12)1. Anal. Calcd. for

(C3H2BO)6•(C9H12)1: C, 63.79; H, 4.77. Found: C, 56.76; H, 4.34. Following guest

S2

removal: Anal. Calcd. for C3H2BO: C, 55.56; H, 3.10. Found: C, 51.26; H, 2.91. Note:

organoboron compounds typically give lowered carbon values in elemental microanalysis

due to the formation of non-combustible boron carbide byproducts.

Synthesis of COF-5. A Pyrex tube measuring o.d. × i.d. = 10 × 8 mm2 was charged with

1,4-benzene diboronic acid (BDBA) (25 mg, 0.15 mmol, Aldrich), 2,3,6,7,10,11-

hexahydroxytriphenylene [(HHTP) 16 mg, 0.050 mmol, TCI] and 1 mL of a 1:1 v:v

solution of mesitylene:dioxane. The tube was flash frozen at 77 K (LN2 bath) and

evacuated to an internal pressure of 150 mtorr and flame sealed. Upon sealing the length

of the tube was reduced to 18 cm. The reaction mixture was heated at 100 ºC for 72 h to

yield a free flowing gray-purple powder. Note that the purple color arises from oxidation

of a small fraction HHTP which exhibits a very large extinction coefficient and is

therefore very highly colored. This side product becomes incorporated within the pores

imparting the purple color to the ‘as synthesized’ form of COF-5. Following guest

removal (see Adsorption Section below) COF-5 is obtained as a light gray solid. Yield:

15 mg, 73 % for C9H4BO2 following guest removal. Anal. Calcld. for C9H4BO2: C,

69.67; H, 2.60. Found: C, 66.48; H, 2.81. Note: organoboron compounds typically give

lowered carbon values in elemental microanalysis due to the formation of non-

combustible boron carbide byproducts. No evidence for the formation of COF-1 was

observed. Note that reaction of BDBA alone at 100 ºC to form COF-1 is slow where after

168 h COF-1 it is obtained in only 25 % yield.

Activation of samples for gas adsorption measurements. COF-1: A 50 mg sample of

COF-1 was heated to 150 oC under dynamic vacuum for 12 h. The sample was back-

filled with nitrogen and then transferred in an air atmosphere to the required vessel for

S3

gas adsorption measurements. COF-5: A 50 mg sample of COF-5 was placed in a 5 mL

glass vial which was subsequently filled with HPLC grade (Aldrich) acetone. After 2

hours of exchange at room temperature the majority of the now yellow-purple acetone

phase was decanted and the vial refreshed with acetone. After 12 hours the solvent was

decanted again and the solid washed with acetone (3 ä 3 mL) and left to air dry in a

desiccator (CaSO4) for 2 hours and then evacuated for 12 h under dynamic vacuum at

ambient temperature. Following evacuation, the sample was back-filled with nitrogen and

then transferred in an air atmosphere to the required vessel for gas adsorption

measurements.

S4

Materials and Methods Section S2: FT-IR Spectroscopy of Starting Materials, Model Compounds, and COFs

FT-IR spectra of starting materials and COFs were obtained as KBr pellets using

Nicolet 400 Impact spectrometer. Assignment and analysis of infrared absorption bands

of starting materials, model compounds, and COF products are presented in this section.

Discussion pertaining to the IR spectral relationships between these compounds is offered

as support for the formation of the covalently linked extended solids.

Figure S1: FT-IR spectrum of benzene 1,4-diboronic acid (BDBA).

S5

Figure S2: FT-IR spectrum of triphenylboroxine (COF-1 model compound).

S6

Figure S3: FT-IR spectrum of COF-1 as synthesized.

S7

Table S1: Peak assignments for FT-IR spectrum of COF-1. Notes are provided to

correlate the spectra of starting material and model compound to that of COF-1.

Peak (cm-1) Assignment and Notes

3429.7 (w) O—H stretch from or end –B(OH)2 groups at the surface of crystallites. 3078.4 (w) Aromatic C—H stretch from phenyl group of COF-1; cf. band from

benzene 1,4-diboronic acid at 3072.3 (w). 3037.6 (w) Aromatic C—H stretch from phenyl group of COF-1; cf. band from

benzene 1,4-diboronic acid at 3047.8 (w) 3017.3 (w) Aromatic C—H stretch from mesitylene guest molecule. Characteristic

methyl group C—H stretching band. Not present in benzene 1,4-diboronic acid starting material nor in FT-IR spectrum of COF-1 following removal of guests.

2915.4 (w) Methyl C—H stretch from mesitylene guest molecule. Characteristic methyl group C—H stretching band. Not present in benzene 1,4-diboronic acid starting material nor in FT-IR spectrum of COF-1 following removal of guests.

2859.4 (w) Methyl C—H stretch from mesitylene guest molecule. Characteristic methyl group C—H stretching band. Not present in benzene 1,4-diboronic acid starting material nor in FT-IR spectrum of COF-1 following removal of guests.

1953.0 (w) C—H overtone band of p-substituted benzene; also observed in spectrum of the benzene 1,4-diboronic acid starting material.

1611.9 (w) Aromatic ring vibration mode (ν8a). Arises only in substituted benzenes and strong in asymmetrically substituted benzenes. Characteristic band. Note absence in benzene 1,4-diboronic acid (Fig. S1). Should be absent in COF-1 since it is a completely symmetric system. It is observed (although weak) due to the population of or end –B(OH)2 groups at the surface of the crystallites in COF-1 which imparts some asymmetry to the system. Band could also be assigned to a sum tone of two low lying fundamentals (e.g. p-xylene).

1515.1 (s) Phenyl ring C=C vibrational mode (ν19a). Characteristic band. Normally strong intensity. Could be overlapped with same band from mesitylene.

1403.1 (s) Can be assigned as phenyl ring C=C vibrational mode (ν19b). Although in the correct region for a p-disubstituted system, its intensity is normally weak. Different position than in triphenyleboroxine model compound, but this is expected because positioning varies according to the substitution pattern of the phenyl ring. Band is also present in the benzene 1,4,-diboronic acid starting material.

1377.7 (s) B—O stretch (characteristic band for boroxine), also present in triphenylboroxine model compound.

1342.0 (s) B—O stretch, shifted by -10 cm-1 from characteristic band for boroxine. 1301.3 (s) C—C stretch, shifted by -10 cm-1 from band observed for

triphenylboroxine model compound.

S8

1260.5 (m) C—B stretch; could be overlapped with C—C stretch. Shifted by -5 cm-1 from band observed for triphenylboroxine model compound, and is the same position as observed in benzene 1,4-diboronic acid starting material.

1174.2 (w) C—H in plane deformation band p-substituted benzene, in the same position observed for benzene 1,4-diboronic acid starting material.

1112.9 (m) C—H in plane deformation band for p-substituted benzene, shifted by -16 cm-1 from band observed for benzene 1,4-diboronic acid starting material.

1087.4 (w) B—C stretch characteristic boroxine compounds, shifted by -5 cm-1 from triphenylboroxine model compound.

1067.0 (w) Possibly B—C stretch or C—H stretch 1026.3 (m) B—C stretch observed in both model compounds and COF-5. 848.1 (w) C—H out of plane deformation band for p-substituted benzene, also

present, but stronger, in benzene 1,4-diboronic acid. 761.5 (m) This peak has previously been assigned in an earlier to a C—H

deformation mode, however the presence of this peak in the exact position in triphenylboroxine (model compound) makes this assignment questionable since these compounds have different aromatic ring systems. Nonetheless, a constant structural element between the two compounds, likely the invariant boroxine ring, accounts for this band.

710.6 (s) Out-of-plane deformation for B3O3 unit. 664.8 (w) Out of plane phenyl ring deformation for p-substituted benzene

S9

Figure S4: Stack plot comparing the FT-IR spectrum of benzene 1,4-diboronic acid (top)

to COF-1 (bottom).

S10

Figure S5: Stack plot comparing the FT-IR spectrum of triphenylboroxine (top) to COF-

1 (bottom).

S11

Figure S6: Stack plot comparing the FT-IR spectrum of COF-1 before (bottom) and

following (top) evacuation of included guests and gas adsorption experiments.

S12

Figure S7: FT-IR spectrum of of 2,3,6,7,10,11-hexahydroxytriphenylene (HHTP).

S13

Figure S8: FT-IR spectrum of 2-Phenyl-1,3,2-benzodioxaborole, model compound for

COF-5.

S14

Figure S9: FT-IR spectrum of ‘as synthesized COF-5.’

S15

Table S2: Peak assignments for FT-IR spectrum of COF-5. Notes are provided to

correlate the spectra of starting materials and model compound to that of COF-5.

Peak (cm-1) Assignment and Notes

3394.1 (m) O—H stretch from or end –B(OH)2 or –OH groups at the surface of crystallites.

3078.4 (w) Aromatic C—H stretch from benzene 1,4-diboronic acid phenyl group of COF-5; cf. band from benzene 1,4-diboronic acid at 3072.3 (w).

2976.5 (w) 2930.7 (w) 2859.4 (w)

C—H stretching from included guest molecules and triphenylene building block.

1637.3 (w) C=C stretch in typical region for fused aromatics. Also present in spectrum of triphenylene.

1525.3 (m) Phenyl ring C=C vibrational mode (ν19a). Characteristic band. Normally strong intensity. Could be overlapped with same band from mesitylene. Shifted by -5 cm-1 from benzene 1,4-diboronic acid.

1494.8 (m) 1448.9 (m)

C=C vibrational modes for triphenylene building block. These are characteristic bands for triphenylene.

1347 (s) B—O stretch (characteristic band for boroxole), shifted by -25 cm-1 from model compound

1332 (s) B—O stretch, shifted by -3 cm-1 from characteristic band for model compound.

1245.3 (s) C—O characteristic stretch for boroxoles; shifted by +5 cm-1 from model compound.

1163.8 (m) 1082.3 (m)

C—H in plane bending modes

1026.3 (m) B—C stretch, also present in model compound 858.3 (m) 832.8 (m)

C—H out of plane bands for p-substituted aromatic.

802.3 (w) C—H out of plane band in region for 1,2,4,5-substituted aromatic 736.1 (w) C—H out of plane band in region for 1,2,4,5-substituted aromatic 664.8 (m) C—H out of plane band shifted by -5 cm-1 from

hexahydroxytriphenylene 613.9 (w) C—H out of plane band shifted by -6 cm-1 from

hexahydroxytriphenylene 542.6 (w) unassigned

S16

Figure S10: Stack plot comparing the FT-IR spectrum of COF-5 (bottom) with benzene

1,4-diboronic acid (BDBA) (top).

S17

Figure S11: Stack plot comparing the FT-IR spectrum of COF-5 (bottom) with

2,3,6,7,10,11-hexahydroxytriphenylene (HHTP) (top).

S18

Figure S12: Stack plot comparing the FT-IR spectrum of COF-5 (bottom) with 2-

Phenyl-1,3,2-benzodioxaborole (top).

S19

Figure S13: Stack plot comparing the FT-IR spectrum of COF-5 before (bottom) and

following (top) removal of included guests by acetone extraction and gas adsorption

experiments.

S20

Materials and Methods Section S3: Scanning Electron Microscopy Imaging (SEM) and Energy Dispersive X-ray (EDX) analysis of COF-1 and COF-5

For SEM imaging Both materials were dispersed over a sticky carbon surface

adhered to a flat aluminum platform sample holder and then gold coated (ambient

temperature, reduced 200 torr pressure in an argon atmosphere, sputtered for 60 s from a

solid gold target at a current at 40 mA). Samples were analyzed using a Hitachi S3200N

Scanning Electron Microscope equipped with Imaging-Everhart-Thornley & Robinson

BSE Detectors and a XEDS - Noran UTW SiLi detector, and was operated in high

vacuum mode using a 30 kV accelerating voltage. Multiple samples of COF-1 and COF-5

were surveyed. Only one unique morphology was apparent after exhaustive examination

of a range of particle sizes that were deposited on the sample holder: clusters of oblong

plates were observed for COF-1 (Figure S14) and piles deformed hexagonal plates

observed for COF-5 (Figure S15). No evidence for the presence of other phases was

observed for either sample. No degradation of either sample was apparent during analysis

which typically lasted 1 - 1.5 h per sample.

For EDX analysis using Hitachi S3200N Scanning Electron Microscope of the

samples were prepared in the same manner as above for SEM imaging excluding gold

coating. Although the elemental compositions of COF-1 and COF-5 consist of atoms

which lie outside the reliable range for quantification by EDX analysis, detection of

(spurious) heaver elements (e.g. Si from reaction vessel) was not observed supporting

that single phase materials have been isolated.

S21

Figure S14: SEM image of COF-1 revealing the clusters oblong plates; scale is inset.

100 µm

Figure S15: SEM image of COF-5 revealing the piles of deformed hexagonal plates;

scale is inset.

25 µm

S22

Materials and Methods Section S4: 11B MAS and 13C CP-MAS Nuclear Magnetic Resonance Studies for COF-1 and COF-5

Data were collected on a Bruker DSX 300 MHz Solid State NMR. Samples were

packed in 5 mm ZrO2 rotors and spun between 5.0 – 8.0 kHz during data collection.

Standard pulse sequences were employed with 1.0 - 1.5 s recycle times found to be

optimal.

Figure S16: Stack plot comparing the 11B NMR spectra of COF-1, triphenylboroxine,

and BDBA diboronic acid. Asterisks (*) indicate peaks arising from spinning side bands.

x106

300

200

100

0

-100

-200

Inte

nsity

-200-1000100200PPM

COF1

triphenylboroxine

*

**

**

*

*

diboronic acid

S23

Figure S17: Stack plot comparing the 11B NMR spectra of COF-5, 2-Phenyl-1,3,2-

benzodioxaborole, and BDBA diboronic acid. Asterisks (*) indicate peaks arising from

spinning side bands.

**

**

**

S24

Figure S18: 13C CP-MAS NMR spectrum of COF-1 (top) vs. triphenylboroxine (bottom)

evidencing the inclusion of mesitylene with observance of signature methyl single at 23.1

ppm. Asterisks (*) indicate peaks arising from spinning side bands.

S25

Materials and Methods Section S5: Structural Models and X-ray Analyses Cerius2 Modeling: Atomic positions in fractional coordinates of unit cell parameters

calculated from Cerius2 were used for model biased Le Bail extractions and illustrations

depicted in manuscript. For both models a single benzene ring was placed at the centroids

of the respective pores of COF-1 (1/3, 2/3, 3/4) and COF-5 (0, 0, 0) to represent guest

molecules. For COF-5 guests were included at 15 % site occupancy to represent the

random population of the mesopores by starting materials, solvent, and byproducts.

Table S3: Fractional atomic coordinates for COF-1 and COF-5 calculated from Cerius2

modeling.

COF-1 COF-5

Hexagonal, P63/mmc a = b = 15.6529, c = 6.7005 Å

Hexagonal, P6/mmm a = b = 30.0198, c = 3.4040 Å

atom x, y, z atom x, y, z B1 0.05772, 0.11543, 0.25000 B1 0.11357, 0.55679, 0.50000 B2 0.44466, 0.72233, 0.25000 O1 0.61305, 0.14429, 0.50000 O1 0.11133, 0.05567, 0.25000 C1 0.94617, 0.47308, 0.50000 O2 0.38900, 0.77800, 0.25000 C2 0.47312, 0.02688, 0.50000 C1 0.11184, 0.38361, 0.25000 C3 0.61950, 0.19328, 0.50000 C2 0.21864, 0.43728, 0.25000 C4 0.66641, 0.23896, 0.50000 C3 0.21837, 0.27685, 0.25000 C5 0.66687, 0.28621, 0.50000 C4 0.11030, 0.88970, 0.25000 C6 0.02886, 0.05722, 0.50000 C5 0.38900, 0.77800, 0.75000 C7 0.05567, 0.02783, 0.50000 C6 0.44466, 0.72233, 0.75000

X-ray Data Collection, Unit Cell Determination, and Le Bail Extraction: Powder X-ray

data were collected using a Bruker D8-Advance θ-2θ diffractometer in reflectance Bragg-

Brentano geometry employing Ni filtered Cu Kα line focused radiation at 1600 W (40

kV, 40 mA) power and equipped with a Na(Tl) scintillation detector fitted a 0.2 mm

S26

radiation entrance slit. Samples were mounted on zero background sample holders by

dropping powders from a wide-blade spatula and then leveling the sample surface with a

razor blade. Given that the particle size of the ‘as synthesized’ samples were already

found to be quite mono-disperse no sample grinding or sieving was used prior to analysis.

The best counting statistics were achieved by collecting samples using a 0.02º 2θ step

scan from 1.5 – 60º with an exposure time of 10 s per step. No peaks could be resolved

from the baseline for 2θ > 50º data and was therefore not considered for further analysis.

Unit cell determinations were carried out using the Powder-X software suite

(PowderX: Windows-95 based program for powder X-ray diffraction data processing", C.

Dong, J. Appl. Crystollogr. (1999), 32, 838) for peak selection and interfacing with the

Treor (TREOR: A Semi-Exhaustive Trial-and-Error Powder Indexing Program for All

Symmetries. Werner, P.-E., Eriksson, L. and Westdahl, M., J. Appl. Crystollogr. 18

(1985) 367) ab inito powder diffraction indexing program. Figure of merits were M10 =

15 for COF-1 and M9 = 18 for COF-5. An internal Si standard (NIST) was used to

normalize peak positions. A low angle calibration of the instrument [using silver

behenate (Gem Dugout) see: Huang T.C., Toraya H., Blanton T.N., Wu Y., J. Appl.

Crystallogr., 1993, 26, 180] was also performed to improve the accuracy of data

collected in this region.

Table S4: Calculated and experimental unit cell parameters for COF-1 and COF-5. Unit cell Parameter Cerius2 Treor Le Bail

COF-1, Hexagonal, P63/mmc a = b (Å) 15.6529 15.056(4) 15.420(1)

c (Å) 6.7005 6.585(3) 6.655(4) COF-5, Hexagonal, P6/mmm

a = b (Å) 30.0198 29.74 (3) 29.70(1) c (Å) 3.4040 3.1(2) 3.460(2)

S27

Le Bail extractions were conducted using the GSAS program using data from 2θ

= 3 – 50º for COF-1 and 2θ = 1.5 to 50º for COF-5. Backgrounds where hand fit with

with six terms applying a shifted Chebyschev Polynomial. Both profiles where calculated

starting with the unit cell parameters indexed from the raw powder patterns and the

atomic positions calculated from Cerius2. Using the model-biased Le Bail algorithm, Fobs

were extracted by first refining peak asymmetry with Gausian peak profiles, followed by

refinement of polarization with peak asymmetry. Unit cells were then refined with peak

asymmetry and polarization resulting in convergent refinements. Once this was achieved

unit cell parameters were refined followed by zero-shift. Refinement of unit cell

parameters, peak asymmetry, polarization and zero-shift were used for the final profiles.

Table S5: Final statistics from Le Bail extractions of COF-1 and COF-5 PXRD data.

COF-1 COF-5

Rp 0.0870 0.0476 wRp 0.1122 0.0635 χ2 10.43 18.46

S28

Figure S19: PXRD pattern of COF-1 (top) compared to patterns calculated from Cerius2

with the stacking of the layers in AB staggered arrangement with P63/mmc space group

symmetry (middle) and (bottom) in AA eclipsed stacking arrangement with P6/mmm.

Note the pattern from the eclipsed model does not match the pattern of COF-1.

3 10 20 30 40 50 60

COF-1

Staggered, matches pattern of COF-1

Eclipsed

S29

Figure S20: PXRD pattern of COF-5 (top) compared to patterns calculated from Cerius2

with the stacking of the layers in AB staggered arrangement with P63/mmc space group

symmetry (middle) and (bottom) in AA eclipsed stacking arrangement with P6/mmm.

Note the pattern from the staggared model does not match the pattern of COF-5.

3 10 20 30 40 50 6

COF-1

Staggered

Eclipsed, matches pattern of COF-5

S30

Figure S21: PXRD patterns of COF-1 following evacuation of guest molecules and gas

adsorption measurements (top) and as synthesized (bottom) illustrating the shifting of

layers which results upon guest removal. Note that the principle 100 and 002 diffraction

peaks of COF-1 are retained.

3 10 20 30 40 50 6

before guest removal and gas sorption (as synthesized)

after guest removal and gas sorption

S31

Figure S22: PXRD patterns of COF-5 following removal of guest molecules by acetone

extraction and gas adsorption measurements (top) and as synthesized (bottom).

3 10 20 30 40 50

before guest removal and gas sorption (as synthesized)

after guest removal and gas sorption

S32

Supplementary Section S6: Low Pressure (0 – 1.0 bar) Gas Adsorption Measurements

for COF-1 and COF-5 at 77, 87, and 293 K

Gas adsorption isotherms were measured volumetrically using a Quantachrome

Autosorb-1 automated adsorption analyzer. A liquid nitrogen bath (77 K) was used for N2

isotherms, an argon bath (87 K) was used for Ar isotherms. Micropore sorption data

using CO2 were collected a 273 K (ice water bath). The N2, Ar, and CO2 gases used were

UHP grade. For measurement of the specific surface areas (As, m2/g) the BET method

was applied. Measured uptakes from Ar isotherms are slightly higher than for N2, we

however report the more conservative N2 data in the manuscript for surface areas and

pore volumes. The higher uptakes for Ar are likely do to its small size which allows more

adatoms to bind into adsorption sites in the frameworks that are too small to

accommodate nitrogen.

For all isotherm plots below closed circles are used for adsorption data points and

open circles are used to indicate desorption data points.

The pore size distribution for COF-1 provided in the manuscript is a composite

histogram NLDFT fitting of CO2 and Ar isotherms. The 0-10 Å segment was taken from

CO2 data and the 10-100 Å segement take from Ar data. They are the valid ranges for

these NLDFT models and a composite figure was generated as the Ar model is not valid

below 10 Å.

S33

Figure S23: Nitrogen gas isotherm for COF-1 measured at 77 K.

0

50

100

150

200

250

300

-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1

P/P0

Upt

ake

(cm

3 g-1

)

S34

Figure S24: Nitrogen gas isotherm for COF-5 measured at 77 K.

0

100

200

300

400

500

600

700

800

-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1

P/P0

Upt

ake

(cm

3 g-1

)

S35

Figure S25: Argon gas isotherm for COF-1 measured at 87 K.

0

50

100

150

200

250

300

350

400

-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1P/P0

Upt

ake

(cm

3 g-1

)

S36

Figure S26: Argon gas isotherm for COF-5 measured at 87 K.

0

200

400

600

800

1000

1200

-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1

P/P0

Upt

ake

(cm

3 g-1)

S37

Figure S27: BET plot for COF-1 calculated from nitrogen adsorption data.

R2 = 0.9999

0.00E+00

2.00E-05

4.00E-05

6.00E-05

8.00E-05

1.00E-04

1.20E-04

1.40E-04

1.60E-04

0 0.02 0.04 0.06 0.08 0.1 0.12P/P0

(P/P

0)/N

(1-P

/P0)

S A = 711 m2 g-1

S38

Figure S28: BET plot for COF-5 calculated from nitrogen adsorption data.

R2 = 0.9946

0.00E+00

5.00E-02

1.00E-01

1.50E-01

2.00E-01

2.50E-01

3.00E-01

0 0.02 0.04 0.06 0.08 0.1 0.12P/P0

1/(W

((P0/P

)-1)

S A = 1590 m2 g-1

S39

Figure S29: BET plot for COF-1 calculated from argon adsorption data.

R2 = 0.9999

0.00E+00

5.00E-02

1.00E-01

1.50E-01

2.00E-01

2.50E-01

3.00E-01

3.50E-01

4.00E-01

4.50E-01

5.00E-01

0 0.02 0.04 0.06 0.08 0.1 0.12P/P0

1/(W

((P0/P

)-1)

S A = 710 m2 g-1

S40

Figure S30: BET plot for COF-5 calculated from argon adsorption data.

R2 = 0.9981

0.00E+00

5.00E-02

1.00E-01

1.50E-01

2.00E-01

2.50E-01

3.00E-01

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14P/P0

1/(W

((P0/P

)-1)

S A = 1723 m2 g-1

S41

Figure S31: De Boer t-plot for COF-1

R2 = 0.9842

780

790

800

810

820

830

840

850

860

870

880

0 2 4 6 8 10 12 14 1De Boer Statistical Thickness (Å)

Upt

ake

(cm

3 g-1

)

6

S42

Figure S32: Carbon dioxide isotherm for COF-1 measured at 273 K used for NLDFT

modeling and pore size distribution calculations. The calculated NLDFT isotherm

(carbon slit pore model) is overlaid as open triangles and fitting error indicated.

0

5

10

15

20

25

30

0 0.005 0.01 0.015 0.02 0.025 0.03

P/P0

Upt

ake

(cm

3 g-1)

Fitting Error = 0.032 %

S43

Figure S33: Argon isotherm for COF-1 measured at 87 K used for NLDFT modeling and

pore size distribution calculations. The calculated NLDFT isotherm (carbon slit pore

model) is overlaid as open triangles and fitting error indicated.

0

50

100

150

200

250

300

350

-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

P/P0

Upt

ake

(cm

3 g-1

)

Fitting Error = 1.01 %

S44

Figure S34: Argon isotherm for COF-5 measured at 87 K used for NLDFT modeling and

pore size distribution calculations. The calculated NLDFT isotherm (silica cylindrical

pore model) is overlaid as open triangles and fitting error indicated.

0

100

200

300

400

500

600

700

800

900

-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

P/P0

Upt

ake

(cm

3 g-1

)

Fitting Error = 0.83 %

S45

Supplementary Section S7: Thermalgravimetry

Samples were run on a TA Instruments Q-500 series thermal gravimetric analyzer

with samples held in platinum pans in an nitrogen atmosphere. A 5 Kmin-1 ramp rate was

used and samples were in tested in their ‘as synthesized’ form following washing

products isolated from reactions with acetone.

Figure S35: TGA trace for COF-1

S46

Figure S36: TGA trace for COF-5.

S47

Supplementary Section S8: Mass spectrum of guests extracted from COF-5 prior to gas

adsorption analysis.

Figure S37: EI-MS spectrum of acetone supernatant from COF-5 activation evidencing

the extraction of BDBC (m/z = 167) and HHTP (m/z = 324.1).

S48