Notes on what's been built, how things work, what broke, and what's coming next. Written by Ky'lin Spears — the only developer on this project.
Added the mole equation calculator to the lab. This was probably the most-requested missing piece — you can now do the full n = m/M calculation interactively, with real molar mass data pulled from the elements database.
The way it works: you enter a chemical formula (or pick an element from the periodic table), and the tool auto-computes the molar mass using real atomic mass values from the elements dataset. Then you can solve for any of the three variables — amount in moles (n), mass in grams (m), or molar mass (M). It shows the triangle breakdown like a textbook would, but computed live.
It also handles multi-atom formulas: H₂O correctly gives 18.015 g/mol, C₆H₁₂O₆ gives 180.158 g/mol, etc. The parser handles subscripts and parentheses for things like Ca(OH)₂.
Dr. Glenn Soltes — my AP Chemistry and Environmental Science teacher, and the STEM Fair coordinator at Suitland — reviewed Atomency and wrote a formal endorsement to my school counselor. He described it as having "meaningful instructional potential" and said he plans to bring it to the science department meeting.
He also gave me a textbook so I could expand the platform's coverage beyond what we use in class. Mr. Adekanye had a separate conversation with me about licensing — mentioned the $10K–$50K range that districts typically budget for this kind of tool, and said the path to formal adoption runs through district-level contact, not school-level.
The assistant principal was told about the project. Said he'd contact the head of the science department. That chain is moving.
None of this changes the mission. But formal district recognition matters for visibility and for the students who'd benefit from teachers actually knowing it exists.
Moved off a subdomain to a dedicated domain. Full redesign of the marketing site. Added PWA support — Atomency now installs as an app on any device and works offline after the first load via a service worker that caches all assets.
The shareable URL system was probably the most technically interesting thing this month. Every molecule and simulation state gets LZ-string compressed and base64 encoded into the URL. You can build a molecule, copy the URL, send it to a student, and they open the exact same canvas state. No server involved — it's all in the URL itself.
This stretch was the heaviest development period. Four interconnected simulation modules landed in about two months.
Acid-Base pH: particle-level dissociation, auto Ka/Kb computation from built-in acid/base database, Henderson-Hasselbalch buffer calculations, strong vs. weak distinction, polyprotic acid handling. The pH value recalculates on every input change — no "submit" step.
Titration Curves: the curve shape changes dynamically as you adjust acid/base strength and concentration. Equivalence point and half-equivalence point are detected and labeled. Indicator color bands overlay the curve. Polyprotic titrations show multiple inflection points correctly.
Gas Laws: all five — Boyle's, Charles', Gay-Lussac's, combined, Ideal. Real-gas Van der Waals correction. Interactive PV diagram. Maxwell-Boltzmann speed distribution rendered as a live curve that shifts with temperature.
Chemical Equilibrium: ICE table solver, Le Chatelier perturbation visualization, Kc/Kp interconversion, reaction quotient (Q) vs K comparison. The equilibrium module directly connects to the equation balancer — balanced equations feed straight into the ICE solver.
Reaction kinetics was the first simulation module beyond the molecular builder. It runs a particle-collision model — actual particles moving in a 2D box, collisions detected geometrically, reaction probability computed from the Arrhenius equation at the current temperature. The activation energy curve and rate constant display update in real time as you slide the temperature.
Nuclear decay uses experimentally measured half-life values from the NUBASE database. Every naturally occurring isotope is in there. Alpha, beta-minus, beta-plus, electron capture, and gamma are all modeled. The decay chain visualization follows through multiple daughter products — you can watch U-238 decay all the way through to Pb-206 step by step.
The molecular engine also got a significant rewrite this period. Bond detection moved from a simple proximity check to a valence-aware system that respects electron counts, orbital hybridization, and octet rules. Formal charge computation was added. This is what makes the molecular property panel actually accurate rather than approximate.
The first version of the molecular builder worked. You could drag carbon, hydrogen, nitrogen, and oxygen onto an SVG canvas, and the system would detect bonds when atoms came within a certain distance. That was it — no properties panel, no molecule name, no export.
But it worked. Bonds formed. You could build water and methane and ammonia. The rendering was rough and there was no undo system, but the core idea was proven: a fully client-side, zero-install molecular builder was technically feasible.
That version never shipped anywhere. It took another six months of work before something worth publishing existed. But that initial build was what made it real — before that it was just an idea.