Values by material NEW

Values gathered for every material by the same rules (two AI judges · decoys · source check) — graphene · rGO (capacitance · volumetric · HER · conductivity · sheet resistance · battery capacity · thermal conductivity) · TMD (HER · capacitance · battery capacity) · MOF (BET · CO₂ · capacitance · drug loading) · COF (BET · capacitance) · phosphorene · black phosphorus (battery capacity). MXene values are on the map — each table compares the same-definition column side by side with MXene.

Status — values were extracted from papers by AI and kept only when two different AI judges agreed the quoted sentence supports them. No domain expert has checked them yet. Randomly chosen passed rows re-read against the papers — MOF·COF: 13 wrong of 200 (removed) · TMD: 0 of 20 · graphene: 0 of 20 · phosphorene: 0 of 20. Please read these as «passed AI judging», not «verified».

The defaults and judging rules here are the current version. When experts reply, or we find better evidence, we will change it — and record the date and the reason on this page.

Defaults we chose — MOF·COF (may change · TMD and graphene: inside each island section)

A value reported only as «surface area» with no model name: for MOFs measured with N₂ at 77 K it is counted as BET (field expert's answer, Oct 27: «almost always BET» · the source sentence is marked «[no model named → read as BET]»). Values measured with CO₂ or argon, values with no gas stated, and COFs are still not counted as BET.
Materials derived from MOFs (e.g. MOF-derived carbons) are excluded (196 rows).
Pore volume is not shown yet — our check found definition errors (micropore volume in the total-pore-volume column); we will add it after fixing the definition.
The framework that links metal units by imine bonds (JNM-3) is marked «MOF/COF hybrid».
CTFs (triazine frameworks) are counted as COFs when the authors call them CTF or COF. Because some argue that high-temperature CTFs are closer to amorphous carbon, they are shown as a separate «CTF» group (Oct 23).
The judges tend to accept «surface area» as BET in papers that use BET. In the second COF round, 3 of 20 sampled rows had only «surface area» without «BET» in the sentence (removed). Some may remain outside the sample.
Sample types: pure · composite (mixed with another material) · modified (guest, doping, degradation) · after use (measured after adsorption/regeneration) · needs check. Default view = pure.

Same name, very different values — half explained

For the same MOF name, BET values differed tens to hundreds of times across papers (pure samples only). On Oct 23 we fixed the name groups, and a large part turned out to be «different materials sharing a name»: MIL-53 ×244 → Al form ×4.7 · Fe form ×7.5 · MOF-74 ×240 → per metal (Ni ×4.6 · Mg ×1.9 …) · ZIF-8 ×152 → ×43 (the ZIF-L phase taken out). The large spreads that remain look like sample and condition differences — UiO-67 (measured after soaking in buffer) · MOF-74(Fe) · MIL-125 (6, 12 and 18 h synthesis). The groups in the table below use the fixed rules.

Name group	Papers	Rows	Lowest	Median	Highest	Spread (×)

BET m²/g

IslandSample

Island	Sample	BET m²/g	Type	Conditions	Source

CO₂ uptake (MOF)

This is the second property of the MOF island. It shows equilibrium uptake from a single-gas (CO₂) isotherm, per mass of sample only. The first judging round (M5) failed. The two AI judges read one line of the judging instructions differently. We added a clarification to that line and re-ran only the judging (M5b). This time it passed (supported 89.5% · wrong 8.6%). We re-read 20 randomly chosen passed rows against the papers and found 0 wrong. For the «CO₂ uptake» rows alone, however, 12.5% were judged wrong. That is above the 10% round threshold. The table below lists only rows that both judges supported. No domain expert has checked them yet.

The mmol/g column is our conversion. We divided mg/g by 44.01 and cm³/g by 22.414 (reading cm³ as STP volume). We did not convert wt%, because papers use different denominators, so that cell is left empty. The value as written is in the next column.
Values measured at different temperatures or pressures cannot be compared. Pick a condition group first.
Isosteric heat, selectivity and breakthrough capacity passed judging but are not shown yet.
Materials made by burning MOFs are excluded (39 rows).

ConditionSample

Sample	mmol/g	As written	Type	Conditions	Source

BET ↔ CO₂ uptake of the same sample

From the 65 CO₂ adsorption papers, we separately extracted the BET of the same samples (M6 · 94.5% passed judging) and paired them. A pair was made only when the paper is the same and the sample name matches to the letter. The pairing rule was fixed before seeing results. One dot = one sample. Values at different conditions cannot be compared, so each condition group is drawn separately.

Condition

How to read: several samples from the same paper appear together, so the dots are not independent. ρ (rank correlation) only summarises the shape and is not evidence (it is not shown below 10 dots). At room temperature and 1 bar, the relation between surface area and CO₂ uptake looks weak. A common reading is that at low pressure «binding sites» (open metal sites, amines) matter more than surface area, but that judgement belongs to domain experts.

Specific capacitance F/g — MOF supercapacitors NEW Oct 28

The MOF island's third property (M8). Values above 3,500 F/g are kept and marked «⚠ > 3,500 F/g» (one specialist's answer · Nov 1). From 172 open MOF supercapacitor papers, 100 drawn at random → 80 with values · two AI judges judged at the same time, sealed from each other, and passed it (833 ● rows · both caught all 20 planted errors). We re-read 20 random passed rows against the papers and found 2 wrong (a two-electrode device value attached to the electrode material name · removed). Only gravimetric (F/g) values are shown: 1177 rows from 135 papers. The property definition is the same as the MXene map's «F/g» column and the graphene island. No domain expert has checked them yet.

MXene ↔ graphene ↔ MOF — specific capacitance F/g on the same definition (median · brackets = number of values · MOF column excludes MOF-derived materials · MXene column = values re-measured with the new engine version on 31 Oct only)

Sample	Electrolyte	MXene	Graphene family	MOF

What you cannot read: «which is the better electrode». The mass basis differs between papers. The comparison leaves out two-electrode device values (three-electrode + not stated only · the table below still shows device values).
Materials made by burning, sulfurizing or oxidizing a MOF are marked «MOF-derived» (523 rows). They are no longer MOFs, so the MOF column above leaves them out.
Values near 5,000 F/g (e.g. Co‐BTB‐LB) are written that way in the paper. They match the source but are above the usual range (~3,500), so read them with care.

Ni- and Co-based MOFs only — compared with MXene on the same definition (median · brackets = number of values · three-electrode + not stated · MOF-derived excluded)

Sample	Electrolyte	MXene	Ni-MOF	Co-MOF	NiCo-MOF

The metal is read from the MOF name (Ni-MOF · Co-BTC · ZIF-67 = Co …). Names that do not show a metal, such as MOF-74, are counted as «other».
Ni and Co MOFs often store charge by redox (battery-type) reactions, so their F/g values come out large. Pure MXene may store charge differently. Do not read «better» from the size of the number alone.
Cells with fewer than ten values come from one or two papers — e.g. the five pure Ni acid values are one Ni-BDC paper (including values in an electrolyte with added KI).

Sample	F/g	Identity · test	Electrolyte · conditions	Source

Drug loading capacity — round D1 (passed AI judging · not yet expert-checked) NEW Nov 1

The MOF island's fourth property (D1 · 194 open papers · 510 values). Only values the paper calls «drug loading capacity» · only those both AI judges backed · 2 wrong in 20 source checks (+ 0 in a second 20). % and mg/g are never converted into each other (what the % is of differs between papers). Each value and its source sentence is on the map under «Explore a property → MOF → drug loading capacity».

Unit	Basis (denominator)	Values	Papers	Median

About half the papers do not state the basis — rows marked «not stated» are not guaranteed to mean the same thing.
The mg/g rows only align μg/mg and g/g by scale (same quantity).
Do not read this as «which MOF carries more drug» — drugs, loading methods and measurements all differ.

MOF drug carriers — particle size (nm), cargo and release trigger, taken from one review

A separately marked layer. 24 MOF nanoparticle drug carriers described in one review: Zhou, Vázquez-González & Willner, Chem. Soc. Rev. 2021, 50, 4541–4563 (doi:10.1039/d0cs01030h). Each value carries the review's sentence. These rows did not go through the two-judge and control rounds, and the original papers (ref numbers in brackets) have not been opened yet — so they are not counted in the map's value total. In «release trigger», gate = a signal that opens a surface lock · degradation = a signal that dissolves the framework itself.

MOF · metal	Particle size nm	Cargo	Release trigger	Original paper · figure	Review sentence
UiO-66-NH₂ Zr	≈330	5-fluorouracil	Zn²⁺ (gate)	[91] · Fig. 1	“amine-modified UiO-66 MOFs with particle sizes of ca. 330 nm were first functionalized by positively charged quaternary ammonium functionalities, followed by the encapsulation of the 5-fluorouracil (Fu) anti-cancer drug”
UMCM-1-NH₂ Zn	≈102	rhodamine 6G (drug model)	acidic pH (gate)	[74] · Fig. 1	“smaller amine-modified UMCM-1 MOFs (size ca. 102 nm) were functionalized with positively charged pyridinium units and loaded with the dye rhodamine 6G acting as a drug model”
PCN-224 + MnO₂ + cell membrane Zr	140.5	apatinib	glutathione (gate)	[98] · Fig. 2	“The MnO2-coated apa-loaded MOF nanoparticles, with an average diameter of 140.5 nm, were further functionalized with the cell membrane extracted from the mouse breast cancer cell line 4T1”
MIL-101-N₃ Fe(III)	100–200	doxorubicin	glutathione / DTT (gate)	[101] · Fig. 3	“Azide-modified Fe(III)-based MOF MIL-101-N3 nanoparticles (size 100–200 nm) were loaded with DOX.”
NU-1000 Zr	≈180 × 70	insulin	phosphate (degradation)	[109] · Fig. 5	“two different Zr-MOFs, NU-1000 [ca. 180 × 70 nm] … and PCN-222 [ca. 210 × 50 nm] … were loaded with insulin”
PCN-222 Zr	≈210 × 50	insulin	phosphate (degradation)	[109] · Fig. 5	“two different Zr-MOFs, NU-1000 [ca. 180 × 70 nm] … and PCN-222 [ca. 210 × 50 nm] … were loaded with insulin”
ZIF-90 Zn	143 ± 3.79	GFP · RNase A-NBC	ATP (degradation)	[112] · Fig. 6	“The ZIF-90 nanoparticles (size 143 ± 3.79 nm) were loaded with proteins such as green fluorescence protein (GFP) or RNase A-NBC exhibiting selective cytotoxicity towards cancer cells”
Cu–Zn(II)TCPP Cu	120	porphyrin linker = photosensitizer	sulfide (degradation)	[116] · Fig. —	“the photosensitizer functions of a Zn(II) porphyrin ligand-based MOF structure (an average size of 120 nm) consisting of Cu2+-ions connected by Zn(II)-tetrakis-(4-carboxyphenyl)porphyrin ligands were fully blocked.”
Zr-DTBA Zr	≈169.4	curcumin	glutathione (degradation)	[122] · Fig. 7	“Redox-triggered release of the anticancer drug curcumin from NMOFs (size ca. 169.4 nm) consisting of 4,4′-dithiobisbenzoic acid (4,4′-DTBA) bridged by Zr was demonstrated”
ZIF-8 Zn	70–300	doxorubicin	pH ≤ 6 (degradation)	[124] · Fig. 8	“The drug-loaded ZIF-8 (size 70–300 nm) revealed cytotoxicity toward different types of breast cancer cells”
ZIF-8 + FA-PEG Zn	≈210	chloroquine	pH ≤ 6 (degradation)	[129] · Fig. 8	“ZIF-8 nanoparticles (diameter ca. 210 nm) were loaded with chloroquine diphosphate (CQ) as autophagy inhibitor”
ZIF-8 Zn	132–167	Zn²⁺-dependent DNAzyme	pH 5.5 (degradation)	[138] · Fig. 9	“ZIF-8 NMOFs (size 132–167 nm) loaded with the Zn2+-ions-dependent DNAzyme strands (1)”
ZIF-8 + PEI Zn	≈280	plasmid DNA (pEGFP-C1)	pH 5.5 (degradation)	[139] · Fig. 10	“the plasmid DNA expressing enhanced green fluorescence protein (PEGFP-C1) was immobilized together with polyethyleneimine (PEI) in the ZIF-8 nanoparticles with an average particle size of ≈280 nm”
ZIF-8 Zn	≈100	CRISPR/Cas9	pH 5–6 (degradation)	[140] · Fig. 11	“In vitro experiments revealed the pH-stimulated release of the CRISPR/Cas9 system at pH = 5.0 and 6.0 from the NMOFs (size ca. 100 nm) due to the degradation of the ZIF-8 nanoparticles”
UiO-AZB (azobenzene) Zr	≈100 (edge)	5-fluorouracil	UV 340 nm (degradation)	[141] · Fig. 12	“The resulting NMOFs show bipyramidal morphology with an average edge length of 100 nm, and were stabilized with a polyethylene glycol coating.”
ZIF-8 + GOx Zn	≈300–350	insulin · anti-VEGF aptamer	glucose → gluconic acid (degradation)	[143] · Fig. 13	“The GOx/insulin-loaded ZIF-8 NMOFs, revealing rhombic dodecahedral structures (ca. 300–350 nm), were suggested as a sense-and-treat insulin release system for controlling diabetes.”
PCN-224 + hyaluronic acid Zr	≈100	doxorubicin	hyaluronidase (gate)	[75] · Fig. 15	“This approach was realized with PCN-224 NMOFs, Zr-based NMOFs (size ca. 100 nm), loaded with DOX, and locked with hyaluronic acid”
UiO-68 + DNA lock Zr	100–130	doxorubicin	VEGF aptamer (gate)	[196] · Fig. 18	“Azide-modified UiO-68 NMOFs (bipyramidal particles 100–130 nm) were reacted with a nucleic acid promoter (7)”
UiO-68 + DNA lock Zr	100–150	apixaban	thrombin aptamer (gate)	[197] · Fig. —	“The UiO-68 NMOFs (bipyramidal particles 100–150 nm) were loaded with the Apixaban blood-clotting drug”
UiO-68 + DNA lock Zr	200–250	doxorubicin	ATP aptamer (gate)	[198] · Fig. 19	“DBCO-modified nucleic acids (9) were covalently linked to the azide-functionalized UiO-68 NMOFs (bipyramidal particles 200–250 nm).”
UiO-68 + DNA lock Zr	250–350	doxorubicin	miRNA-21 / miRNA-221 + Exo III (gate)	[200] · Fig. 20	“the preparation of DOX-loaded UiO-68 NMOFs (bipyramidal particles 250–350 nm) locked with duplex nucleic acids”
UiO-68 + DNA lock Zr	250–300	doxorubicin	ATP-activated DNAzyme (gate)	[201] · Fig. 21	“the activation of the DNAzyme cleavage of duplex-locked NMOFs (bipyramidal particles 250–300 nm)”
UiO-68 + DNA lock Zr	100–150	camptothecin	nucleases (gate)	[204] · Fig. 22	“The nucleic acid (18)/(20)/(22)/(24)-functionalized NMOFs (bipyramidal particles 100–150 nm) were loaded with anti-cancer drug, camptothecin, CPT.”
UiO-68 + DNA hydrogel Zr	280–350	doxorubicin	ATP (hydrogel gate)	[205] · Fig. 23	“The azide-modified UiO-68 NMOFs (bipyramidal particles 280–350 nm) were modified with the DBCO-functionalized nucleic acid (27) through a triazine bridge.”

Sizes are the numbers in the review sentence («ca.» = ≈ · two numbers with × = length × width of rod-shaped particles · for UiO-AZB, the edge length of the bipyramid).
Most carriers are Zr (UiO · PCN · NU) or Zn (ZIF) frameworks. The review rarely gives drug loading (wt%) — that needs the original papers.
Results such as cancer-versus-normal cell death are not put side by side, because cells and times differ from paper to paper.

TMD — MoS₂ hydrogen evolution (HER)

It starts from MoS₂ researchers' own papers (76 open papers · the search neither required nor excluded MXene). The property definitions (η10 = overpotential at 10 mA cm⁻² · Tafel slope) are word for word the same as the MXene map's «catalysis (HER)» columns, so the two can be compared on one scale below. Only values both AI judges supported are shown (96.3% · 20/20 in the post-pass sample). No domain expert has checked them yet.

MXene ↔ MoS₂ on the same definition (median · brackets = number of values)

Property	Sample	Electrolyte	MXene	MoS₂

What you can read: pure MXene is a weak HER catalyst in acid. As composites, both drop a lot (get better). For overpotential and Tafel slope, lower is better.
What you should not read: «MXene composites beat MoS₂ composites» — the partners differ (many MXene composites contain Pt or MoS₂), and conditions (concentration, loading, iR correction) are not matched. Cells with about ten values are small.
«Pure» = no partner material listed (doping is not counted as a partner). Phase (1T · 2H) is read only when the paper writes it in the composition — many are «not stated».

Sample	Value	Phase · role	Electrolyte · conditions	Source

Graphene — graphene · rGO hydrogen evolution (HER)

It starts from graphene researchers' own papers (80 papers drawn at random from 223 open papers). The property definitions are word for word those of the MXene map's «catalysis (HER)» and the TMD island. The column only this island has is «what is this sample, really?» — graphene · rGO (reduced graphene oxide) · GO · doped graphene · graphite, as the paper names it, checked against the source by two AI judges (supported 87.8% · post-pass sample 20/20). No value has been checked by an expert yet.

↓ Second property — specific capacitance F/g (supercapacitors · 3,327 values · 479 papers)
↓ Third property — electrical conductivity · sheet resistance · resistivity (new Oct 27)

MXene ↔ graphene on the same definition (median · brackets = number of values)

Property	Sample	Electrolyte	MXene	Graphene family

What you can read: catalysts made only of graphene-family carbon have a higher acid η10 median than even pure MXene (weaker). Most values are the scores of other catalysts (MoS₂ · metals · phosphides) that graphene supports — in the role reading (two AI judges · 114 samples) graphene itself was the active site (protagonist) in only 9 composites; mostly «support» or «conductive aid».
What you should not read: «graphene composites are worse than MXene composites» — partners, electrode supports and loadings differ.
Limit of the identity column: the judges missed 3 rows that wrote «NrGO» (N-doped rGO) as «N-doped graphene» — trust the graphene/rGO split inside the doped group less.

Sample	Value	Identity · role	Electrolyte · conditions	Source

Specific capacitance F/g — graphene-family supercapacitors

Values above 3,500 F/g are kept and marked «⚠ > 3,500 F/g» (one specialist's answer · Nov 1). The graphene island's second property, from two rounds on open graphene-family supercapacitor papers — G2 (120 of 942 drawn at random · 981 ● rows · 20 random rows re-checked against the source, 0 wrong) and G4 (Oct 31 · the remaining 567 papers · both judges 91.6% · ✕ 3.8% · 0 wrong in 20 re-checked rows). Two AI judges worked at the same time, sealed from each other, and both rounds passed. Now 3,327 values from 479 papers. The property definition is word for word the same as the MXene map's «F/g» column. Only gravimetric (F/g) values are shown here — areal, volumetric and retention values were recorded only.

MXene ↔ graphene — specific capacitance F/g on the same definition (median · brackets = number of values · MXene column = values re-measured with the new engine version on 31 Oct only)

Sample	Electrolyte	MXene	Graphene family

What you cannot read: «which is the better electrode». The comparison leaves out two-electrode device values (usually much smaller) — three-electrode and not-stated only. The mass basis (active material, whole electrode) differs between papers. Most composite values are the performance of another material (oxide, polymer) supported by graphene.
Electrolytes were grouped from the condition text (acid = H2SO4 etc. · alkaline = KOH etc. · neutral = Na2SO4 etc.).

Sample	F/g	Identity · test	Electrolyte · conditions	Source

Electrical conductivity · sheet resistance · resistivity — graphene family

The graphene island's third property (G3). From 1,362 open graphene-family conductivity / sheet-resistance papers, 100 drawn at random → 44 with values · two AI judges judged at the same time, sealed from each other, and passed it (186 of 197 rows ● · 1.5% wrong · 20 random rows re-checked against the source, 1 wrong — a high-surface-area graphite written as graphene, removed). The three are kept apart: conductivity is shown in S/cm, sheet resistance (Ω/sq) and resistivity (Ω·cm) separately. We did not convert between them (most values have no thickness).

MXene ↔ graphene — electrical conductivity S/cm, median (brackets = number of values)

Sample	MXene	Graphene family

What you cannot read: «MXene conducts better than graphene». Pure MXene values are mostly well-made free-standing films; the graphene side mixes pressed pellets, aerogels, inks and small amounts inside polymers. Thickness, method and substrate were not matched.
Most «composite» values are a little graphene inside a polymer or oxide, so they are not a property of graphene itself.

Sample	Value (common unit)	Identity	Conditions	Source

Phosphorene · black phosphorus — black-phosphorus battery anode capacity

It starts from black-phosphorus researchers' own battery papers — all 36 open-access papers (a small island · no sampling). The property definition is word for word the same as the MXene map's «battery mAh/g» column. This island's own column is «what is this sample really» — phosphorene (mono- or few-layer), exfoliated black-phosphorus nanosheets, or unexfoliated bulk black phosphorus. Two AI judges judged at the same time, each sealed off so neither could see the other's results (the first round run this way). The 9 samples written as phosphorene were re-asked for identity only; for all 9, both judges answered «phosphorene».

MXene ↔ black phosphorus on the same definition (half cells only · median · brackets = number of values)

Sample	Ion	MXene	Black-phosphorus family

What you can read: black phosphorus has a large theoretical capacity (about 2,600 mAh/g), so its values come out higher than MXene's — the lithium composite median is about twice MXene's.
What you cannot read: «black phosphorus is the better electrode». Most values are early cycles or short tests, and black phosphorus fades fast with cycling (retention is not in this column). There is only one pure black-phosphorus lithium value, and only three sodium values on the black-phosphorus side.
Left out: full-cell values (may be per counter-electrode mass · shown in the table but not compared) · areal (mAh/cm²) values · lithium-sulfur values per sulfur mass.

Sample	mAh/g	Identity · cell	Conditions	Source

TMD — MoS₂-family battery capacity (mAh/g) NEW Nov 1

The TMD island's second property. It starts from TMD researchers' own battery papers (MoS₂, WS₂ …) — 64 open-access papers, 42 with values. The property definition is word for word the same as the MXene map's «battery mAh/g» column and the black-phosphorus island. Two AI judges judged at the same time, each sealed off from the other (passed 97.4% · wrong 1.2%), and 20 passed rows were re-checked against the source text: 1 was wrong. That row recorded as MoS₂ a sample the paper calls «MoSx (close to MoS₃)», so all values of that paper were removed (8 values).

MXene ↔ TMD on the same definition (half cells only · median · brackets = number of values)

Sample	Ion	MXene	TMD family

What you can read: on the same definition (half cell · per mass) the TMD median is higher than MXene's — about three times for pure lithium samples.
What you cannot read: «TMDs are the better electrode». Most values are early cycles or values after a set number of cycles, and current density and mass basis differ between papers (retention is not in this column). The phase column only reads what the paper writes in the name — «unknown» is the largest group.
Left out: full-cell values (shown, not compared) · areal (mAh/cm²) · lithium-sulfur values per sulfur mass · the MoSx paper whose identity needs checking.

Sample	mAh/g	Phase · cell	Conditions	Source

How it was made — for each round, randomly chosen papers (MOF 664 · COF 416 · MoS₂ 76 · graphene 80) were extracted with rules written down in advance → two judges · the judges themselves are tested with planted wrong values · 20 random passed rows re-checked afterwards. Same method as the main map.

If a value looks wrong, please post on the Q&A page with its DOI so we can find it.