Ag-TiO2 on cotton: band gap and dye degradation kinetics checked

Silver-TiO2 on cotton fabric breaks down a blue dye under UV light. We check its 3.55 eV indirect gap and first-order rate with fibub's Tauc plot and kinetics tools.

Cotton fabric coated with tiny particles of titanium dioxide (TiO2) and silver can clean water by itself when light shines on it: the particles soak up ultraviolet light and use its energy to break down dyes. A team at the National Autonomous University of Mexico (UNAM) made such a fabric and tested it under six light sources. This article shows how the tauc plot turns a reflectance spectrum into the material’s band gap, and how the reaction kinetics fitter turns “how much dye is left” readings into a rate constant you can compare between lamps.

The paper in brief

The team grew TiO2 (anatase, the most active form for this job) and silver nanoparticles on cotton. Chemical mapping found about 1.4 % silver and 1.4 % titanium by weight; electron microscope pictures showed TiO2 grains of about 10 nm and silver particles of about 21 nm.

The band gap is the smallest packet of light energy the material can absorb to free an electron, the first step in breaking down a dye. From a Tauc plot of the fabric’s spectrum, read as an indirect transition (one where the electron also needs a nudge from a lattice vibration), they found 3.55 eV, higher than the 3.2 eV of their plain anatase reference. That means the fabric needs light shorter than roughly 350 to 360 nm: ultraviolet.

They then soaked the fabric in a blue dye (the one they track absorbs most strongly at 672 nm) and lit it. Under a mercury lamp about 88 % of the dye was gone after 90 minutes; under a strong blue LED only about 33 %; bare cotton with no catalyst removed under 5 %. Lamps rich in UV worked well, and sunlight did about as well as the mercury lamp. The paper’s main point is that the light source’s spectrum has to be built into the kinetic model, not just its total power.

Checked with the tools

We did two things, and we say which is which. The band gap step is a simulation: we made up a spectrum with the paper’s 3.55 eV indirect gap and checked that the Tauc plot tool’s maths finds it again. The kinetics step is a recomputation from the paper’s stated end points (88 %, 33 %, 5 % at 90 min), plus simulated readings every 10 minutes to show the fit. Neither uses the paper’s measured spectra or data tables.

Step 1: the band gap. We built a diffuse-reflectance spectrum from 250 to 500 nm with an indirect edge at 3.55 eV, a 60 meV soft tail below it and a little noise, then ran it through the tool’s Kubelka-Munk conversion and its automatic straight-line search with n = ½. The tool found Eg = 3.550 ± 0.0002 eV (Figure 1), which is 349 nm. A plain anatase spectrum built the same way with 3.2 eV came back as 3.200 eV (388 nm). So the 0.35 eV shift the paper reports moves the fabric’s absorption edge about 38 nm deeper into the UV.

Tauc plot, indirect transition (simulated Ag-TiO2 spectrum)Tauc plot, indirect transition (simulated Ag-TiO2 spectrum)01233.544.55Photon energy (eV)(F(R)·hν)^½Eg = 3.55 eVSpectrumFitted straight linefibub.com
Figure 1. A spectrum we simulated with an indirect gap of 3.55 eV (the paper's value) and a 60 meV tail, read as diffuse reflectance and fitted by the Tauc plot tool's maths. Not the paper's measured spectrum.

Step 2: the transition type matters. Reading the very same simulated spectrum as a direct transition (n = 2) gave 3.16 eV, about 0.4 eV off. The tool cannot tell you which transition is right; that comes from the material (anatase is usually treated as indirect, as the paper does). Always report which one you used.

Step 3: the rate constants. If dye loss is first order (the rate is proportional to how much dye is left, the usual starting model for dilute dyes), then C/C0 = e^(-kt) and k = -ln(C/C0) / t. From the paper’s end points:

Light Dye gone at 90 min (paper) k (per min, our recomputation) Half-life
Mercury lamp about 88 % 0.0236 29 min
Blue LED about 33 % 0.0045 156 min
No catalyst under 5 % under 0.00057 over 20 h

The mercury lamp run is about 5 times faster than the blue LED, even though the LED is far brighter in total (the paper gives about 2,360 W/m² for it). That fits the 3.55 eV gap: almost none of the LED’s 450 nm light carries enough energy.

Step 4: fitting the curve. We simulated readings every 10 minutes from those k values with 1.5 % noise and fitted them with the kinetics fitter (Figure 2). For the mercury lamp run it gave back k = 0.0236 ± 0.0001 per min, R² = 0.9999, and first order won the model comparison outright. For the blue LED it gave 0.0044 ± 0.0001 per min, with first order preferred (87 % Akaike weight). For the no-catalyst run the dye barely moves, the noise is as big as the change, and the tool cannot say which order fits: a useful warning that a near-flat control should be reported as “no measurable reaction”, not as a rate.

Dye left over time (simulated runs, first-order fits)Dye left over time (simulated runs, first-order fits)00.250.50.751020406080Time (min)C/C0Mercury lamp, 88 % in 90 minBlue LED, 33 % in 90 minNo catalyst, under 5 %fibub.com
Figure 2. Runs we simulated from the paper's 90-minute end points (dots, 1.5 % noise) and the Reaction kinetics fitter's first-order fits (lines). An illustration, not the paper's data.

How we made the graphs

The script scripts/blog-figures/ag-tio2-cotton-photocatalyst-band-gap-dye-kinetics.mjs makes both charts and writes them into this page. It calls exampleAlpha, toSpectrum, taucY, autoWindow and gapFit from src/lib/sci/tauc.ts (the maths of the Tauc plot tool) and conc, fitOrders and akaikeWeights from src/lib/sci/kinetics.ts (the maths of the Reaction kinetics fitter). From the paper: the gaps (3.55 and 3.2 eV) and the 90-minute end points. Our own assumptions: the 60 meV tail, the absorption strength, the noise levels, the 10-minute sampling and first-order kinetics.

How we test the maths

Every change to the site runs node tests/sci.mjs, which includes these checks:

  • tests/sci-tools/tauc-plot.mjs: an exact indirect-gap spectrum built with a 1.12 eV gap must give back 1.12000 eV with n = ½; a direct-gap spectrum read as diffuse reflectance through Kubelka-Munk must give the same 1.73 eV as when read as absorption; the page’s example must match a NumPy fit (2.399323 ± 0.000123 eV).
  • tests/sci-tools/reaction-kinetics-fitter.mjs (added with this article): 88 % gone in 90 min must mean k = 0.0235585 per min; exact readings every 10 min must fit back to that k with a 29.42 min half-life, with first order preferred.
  • tests/sci.mjs (kinetics block): exact zero-, first- and second-order data must each return their own k and win the comparison, and the textbook methyl isonitrile Arrhenius data must give an activation energy near 160 kJ/mol.

Check it yourself

On the tauc plot:

  1. Paste these ten lines into Two columns: wavelength (or energy), then the reading: 3.3 0, 3.4 0, 3.5 0, 3.6 6.9, 3.7 60.8, 3.8 164.5, 3.9 314.1, 4 506.3, 4.1 737.8, 4.2 1006 (one pair per line).
  2. Set First column to Photon energy (eV) and Second column to Absorption coefficient α (cm⁻¹).
  3. Set Transition to Indirect allowed, n = ½ and leave Choose the region on Automatically.
  4. The result should read an indirect gap of about 3.55 eV (349 nm). Switch Transition to Direct allowed, n = 2 and watch the answer move.

On the reaction kinetics fitter:

  1. Set Time unit to min and type C/C0 in Concentration unit.
  2. Paste into Time and concentration, one pair per line: 0 1, 15 0.702, 30 0.493, 45 0.346, 60 0.243, 75 0.171, 90 0.120.
  3. The headline should say First order fits best, with k close to 0.0236 1/min (per minute), half-life about 29.4 min. Press ln [A] to see the straight line that proves it.

What researchers can use it for

  • Reporting band gaps of coated fabrics, films and powders: the tauc plot takes diffuse reflectance directly, gives the gap with a standard error and shows how much it moves with the chosen window.
  • Comparing light sources fairly: fit each run with the reaction kinetics fitter and compare k values rather than “percent removed at 90 min”, which depends on when you stop.
  • Spotting a control with no real reaction: a flat run gives a k with an error as large as itself and no clear best order.
  • Temperature studies: the kinetics fitter’s Arrhenius box turns k at several temperatures into an activation energy.
  • Teaching: the two tools together show students why a bright lamp can still be the slow one when its photons are below the band gap.

Sources

  • Alvarez-Amparán, M. A., Chacon-Argaez, U. and Cedeño-Caero, L. (2026), The Effect of Electronic and Optical Properties on the Kinetic Photocatalytic Model of Methyl Blue Degradation, Molecules 31(5), 782, doi:10.3390/molecules31050782. Open access under a Creative Commons Attribution licence (the page metadata links CC BY 3.0), checked on the journal page, 2026-11-16. Values used here: indirect band gap 3.55 eV for the Ag-TiO2 cotton (AgTiC); 3.2 eV for the anatase reference and 3.0 eV for Degussa P25; about 360 nm minimum activation wavelength; 1.38 wt% Ag and 1.36 wt% Ti (EDS); Ag particles about 21 nm and TiO2 about 9.6 nm (TEM); dye absorption peak 672 nm; about 88 % of the dye gone after 90 min under the mercury lamp, about 33 % under the blue LED, under 5 % on bare cotton.
  • Tauc plot method: Tauc, J., Grigorovici, R. and Vancu, A. (1966), Phys. Status Solidi 15, 627; Kubelka-Munk function for diffuse reflectance (the formulas the Tauc plot tool uses).

If you are an author or publisher and want a credit or a passage changed or removed, tell us and we will act quickly.

Tools in this article

Read next