Q-props

Q-props is designed to deliver the world’s most accurate predictions of physical properties for pure components and mixtures

Problem: Garbage in, garbage out

Computer simulations are playing an increasingly important role in process and product development but rely fully on the accuracy of the physical property data used. As experimental data is often unavailable or unreliable, accurately predicting physical properties of components and mixtures is typically the most critical part of a good simulation.

Due to their ease of use and low computational cost, group contribution methods (e.g. UNIFAC) are in widespread use despite their well-documented lack of accuracy across application areas. Quantum chemical methods have gained traction but continue to be limited by the need for computational power and expert users. Also, no practically applicable quantum method consistently yields predictions within “chemical accuracy” (<1 kcal/mol difference from prediction in Gibbs energy or enthalpy to experimental data of formation properties or equilibrium constants) and they give no indication whether a specific prediction is reliable, even though errors can be >10 kcal/mol.

Solution: Q-props

Q-props is the first development in years to drastically improve the accuracy and reliability of property predictions from quantum chemical methods. Q-props aims to consistently deliver chemical accuracy (<1 kcal/mol) and always returns a prediction-specific uncertainty, so you know how much you can trust your simulation. Q-props is also easy to use; it will interface with leading process simulation tools, performs heavy calculations in the cloud and handles all expert choices.

Q-props achieves its prediction accuracy by combining experimental data and quantum calculations for components similar to the target molecule. A wide array of quantum descriptors are used to form error-cancelling reaction networks and a sophisticated AI is applied to decide what experimental data, reactions and methods to use. The AI is optimized for highest possible accuracy while keeping all information relevant to get the prediction specific uncertainty.

Example: Heat of formation for ethylbenzene and tert-butanol

The table below shows the heat of formation of ethylbenzene and tert-butanol as predicted using a group contribution method (Joback), quantum chemistry (G3MP2), and Q-props, as well as the three experimental data points available from NIST:

Method ΔfH, 25oC [kJ/mol]
Ethylbenzene Tert-butanol
Group Contribution 28.1 -286.9
G3MP2 23.1 -313.3
Q-props 29.5±3.2 -313.8±3.2
NIST Webbook 29.8±0.8,
49.0±4.0,
69.3
-312.6±0.9,
-313.0±1.5,
-309.7

Ethylbenzene

Tert-butanol

Of the three predictions, Q-props is consistently close to the experimental data points from NIST. Group contribution also yields chemical accuracy in the case of ethylbenzene, but is far off in the case of tert-butanol, whereas the quantum method G3MP2 gives better results in the case of tert-butanol than for ethylbenzene. Notably, only Q-props gives an uncertainty estimate. This derives directly from the reaction network method visualized on the figures above.

In this case, intelligently using quantum chemistry and experimental data of similar compounds via a network of approximately 2000 (ethylbenzene) and 400 (tert-butanol) self-selected reactions, Q-props gives the most accurate prediction, a specific uncertainty, and even a very good indication of what experimental data to trust, when such data is available but contradictory.