Mixtures Recomposition by Neural Nets: A Multidisciplinary Overview.

Artificial Neural Networks (ANNs) are transforming how we understand chemical mixtures, providing an expressive view of the chemical space and multiscale processes. Their hybridization with physical knowledge can bridge the gap between predictivity and understanding of the underlying processes. This overview explores recent progress in ANNs, particularly their potential in the 'recomposition' of chemical mixtures. Graph-based representations reveal patterns among mixture components, and deep learning models excel in capturing complexity and symmetries when compared to traditional Quantitative Structure-Property Relationship models. Key components, such as Hamiltonian networks and convolution operations, play a central role in representing multiscale mixtures. The integration of ANNs with Chemical Reaction Networks and Physics-Informed Neural Networks for inverse chemical kinetic problems is also examined. The combination of sensors with ANNs shows promise in optical and biomimetic applications. A common ground is identified in the context of statistical physics, where ANN-based methods iteratively adapt their models by blending their initial states with training data. The concept of mixture recomposition unveils a reciprocal inspiration between ANNs and reactive mixtures, highlighting learning behaviors influenced by the training environment.

Artificial intelligence-driven design of fuel mixtures.

High-performance fuel design is imperative to achieve cleaner burning and high-efficiency engine systems. We introduce a data-driven artificial intelligence (AI) framework to design liquid fuels exhibiting tailor-made properties for combustion engine applications to improve efficiency and lower carbon emissions. The fuel design approach is a constrained optimization task integrating two parts: (i) a deep learning (DL) model to predict the properties of pure components and mixtures and (ii) search algorithms to efficiently navigate in the chemical space. Our approach presents the mixture-hidden vector as a linear combination of each single component's vectors in each blend and incorporates it into the network architecture (the mixing operator (MO)). We demonstrate that the DL model exhibits similar accuracy as competing computational techniques in predicting the properties for pure components, while the search tool can generate multiple candidate fuel mixtures. The integrated framework was evaluated to showcase the design of high-octane and low-sooting tendency fuel that is subject to gasoline specification constraints. This AI fuel design methodology enables rapidly developing fuel formulations to optimize engine efficiency and lower emissions.

Temperature and Pressure Dependent Rate Coefficients for the Reaction of Ketene with Hydroxyl Radical.

The reaction of ketene with hydroxyl radical is drawing growing attention, for it is found to constitute an important step during the combustion of hydrocarbon and oxygenated hydrocarbon fuels, e.g., acetylene, propyne, allene, acetone, gasoline, diesel, jet fuels, and biofuels. We studied the potential energy surface (PES) of this reaction using B2PLYP-D3/cc-PVTZ for geometry optimization and composite methods based on CCSD(T)-F12/cc-PVTZ-F12 for energy calculations. From this PES, temperature- and pressure-dependent rate coefficients and branching ratios at 200-3000 K and 0.01-100 atm were derived using the RRKM/ME approach. The reaction is dominated by four product channels: (i) OH addition on the olefinic carbon of ketene to form CH2OH + CO, which is the most dominant under all conditions; (ii) H abstraction producing HCCO + H2O, which is favored at high temperatures; (iii) OH addition on the carbonyl carbon to form CH3 + CO2, which is favored at low pressures and high temperatures; and (iv) collisional stabilization of CH2COOH, which is favored at high pressures and low temperatures. With increasing temperatures, the overall rate constant koverall exhibit first negative but then positive temperature dependency, with its switching point (also the minimum point) at ∼400 K. Both product channel CH2OH + CO and HCCO + H2O are independent of pressure, whereas formation of CH3 + CO2 and collisional stabilization of CH2COOH are highly pressure dependent. Fitted modified Arrhenius expressions of the calculated rate constants are provided for the purpose of combustion modeling.

Absence of detectable influenza RNA transmitted via aerosol during various human respiratory activities--experiments from Singapore and Hong Kong.

Two independent studies by two separate research teams (from Hong Kong and Singapore) failed to detect any influenza RNA landing on, or inhaled by, a life-like, human manikin target, after exposure to naturally influenza-infected volunteers. For the Hong Kong experiments, 9 influenza-infected volunteers were recruited to breathe, talk/count and cough, from 0.1 m and 0.5 m distance, onto a mouth-breathing manikin. Aerosolised droplets exhaled from the volunteers and entering the manikin's mouth were collected with PTFE filters and an aerosol sampler, in separate experiments. Virus detection was performed using an in-house influenza RNA reverse-transcription polymerase chain reaction (RT-PCR) assay. No influenza RNA was detected from any of the PTFE filters or air samples. For the Singapore experiments, 6 influenza-infected volunteers were asked to breathe (nasal/mouth breathing), talk (counting in English/second language), cough (from 1 m/0.1 m away) and laugh, onto a thermal, breathing manikin. The manikin's face was swabbed at specific points (around both eyes, the nostrils and the mouth) before and after exposure to each of these respiratory activities, and was cleaned between each activity with medical grade alcohol swabs. Shadowgraph imaging was used to record the generation of these respiratory aerosols from the infected volunteers and their impact onto the target manikin. No influenza RNA was detected from any of these swabs with either team's in-house diagnostic influenza assays. All the influenza-infected volunteers had diagnostic swabs taken at recruitment that confirmed influenza (A/H1, A/H3 or B) infection with high viral loads, ranging from 10(5)-10(8) copies/mL (Hong Kong volunteers/assay) and 10(4)-10(7) copies/mL influenza viral RNA (Singapore volunteers/assay). These findings suggest that influenza RNA may not be readily transmitted from naturally-infected human source to susceptible recipients via these natural respiratory activities, within these exposure time-frames. Various reasons are discussed in an attempt to explain these findings.

Different types of door-opening motions as contributing factors to containment failures in hospital isolation rooms.

Hospital isolation rooms are vital for the containment (when under negative pressure) of patients with, or the protection (when under positive pressure) of patients, from airborne infectious agents. Such facilities were essential for the management of highly contagious patients during the 2003 severe acute respiratory syndrome (SARS) outbreaks and the more recent 2009 A/H1N1 influenza pandemic. Many different types of door designs are used in the construction of such isolation rooms, which may be related to the space available and affordability. Using colored food dye as a tracer, the qualitative effects of door-opening motions on the dissemination of potentially contaminated air into and out of a single isolation room were visualized and filmed using Reynolds-number-equivalent, small-scale, water-tank models fitted with programmable door-opening and moving human figure motions. Careful scaling considerations involved in the design and construction of these water-tank models enabled these results to be accurately extrapolated to the full-scale situation. Four simple types of door design were tested: variable speed single and double, sliding and hinged doors, in combination with the moving human figure. The resulting video footage was edited, synchronized and presented in a series of split-screen formats. From these experiments, it is clear that double-hinged doors pose the greatest risk of leakage into or out of the room, followed by (in order of decreasing risk) single-hinged, double-sliding and single-sliding doors. The relative effect of the moving human figure on spreading any potential contamination was greatest with the sliding doors, as the bulk airflows induced were large relative to those resulting from these door-opening motions. However, with the hinged doors, the airflows induced by these door-opening motions were significantly greater. Further experiments involving a simulated ventilated environment are required, but from these findings alone, it appears that sliding-doors are far more effective for hospital isolation room containment.

Airflow dynamics of human jets: sneezing and breathing - potential sources of infectious aerosols.

Natural human exhalation flows such as coughing, sneezing and breathing can be considered as 'jet-like' airflows in the sense that they are produced from a single source in a single exhalation effort, with a relatively symmetrical, conical geometry. Although coughing and sneezing have garnered much attention as potential, explosive sources of infectious aerosols, these are relatively rare events during daily life, whereas breathing is necessary for life and is performed continuously. Real-time shadowgraph imaging was used to visualise and capture high-speed images of healthy volunteers sneezing and breathing (through the nose - nasally, and through the mouth - orally). Six volunteers, who were able to respond to the pepper sneeze stimulus, were recruited for the sneezing experiments (2 women: 27.5±6.36 years; 4 men: 29.25±10.53 years). The maximum visible distance over which the sneeze plumes (or puffs) travelled was 0.6 m, the maximum sneeze velocity derived from these measured distances was 4.5 m/s. The maximum 2-dimensional (2-D) area of dissemination of these sneezes was 0.2 m(2). The corresponding derived parameter, the maximum 2-D area expansion rate of these sneezes was 2 m(2)/s. For nasal breathing, the maximum propagation distance and derived velocity were 0.6 m and 1.4 m/s, respectively. The maximum 2-D area of dissemination and derived expansion rate were 0.11 m(2) and 0.16 m(2)/s, respectively. Similarly, for mouth breathing, the maximum propagation distance and derived velocity were 0.8 m and 1.3 m/s, respectively. The maximum 2-D area of dissemination and derived expansion rate were 0.18 m(2) and 0.17 m(2)/s, respectively. Surprisingly, a comparison of the maximum exit velocities of sneezing reported here with those obtained from coughing (published previously) demonstrated that they are relatively similar, and not extremely high. This is in contrast with some earlier estimates of sneeze velocities, and some reasons for this difference are discussed.

A molecular approach for unraveling surface phase transitions: sulfation of BaO as a model NO(x) trap.

SO(3) -induced surface reconstruction: The SO(3) molecule as a multidentate ligand induces remarkable surface reconstruction phenomena on alkaline earth oxide surface. By using ab initio computations, adsorption properties are derived to elucidate the thermodynamics of the SO(3) -BaO system.

Airflow dynamics of coughing in healthy human volunteers by shadowgraph imaging: an aid to aerosol infection control.

Cough airflow dynamics have been previously studied using a variety of experimental methods. In this study, real-time, non-invasive shadowgraph imaging was applied to obtain additional analyses of cough airflows produced by healthy volunteers. Twenty healthy volunteers (10 women, mean age 32.2±12.9 years; 10 men, mean age 25.3±2.5 years) were asked to cough freely, then into their sleeves (as per current US CDC recommendations) in this study to analyze cough airflow dynamics. For the 10 females (cases 1-10), their maximum detectable cough propagation distances ranged from 0.16-0.55 m, with maximum derived velocities of 2.2-5.0 m/s, and their maximum detectable 2-D projected areas ranged from 0.010-0.11 m(2), with maximum derived expansion rates of 0.15-0.55 m(2)/s. For the 10 males (cases 11-20), their maximum detectable cough propagation distances ranged from 0.31-0.64 m, with maximum derived velocities of 3.2-14 m/s, and their maximum detectable 2-D projected areas ranged from 0.04-0.14 m(2), with maximum derived expansion rates of 0.25-1.4 m(2)/s. These peak velocities were measured when the visibility of the exhaled airflows was optimal and compare favorably with those reported previously using other methods, and may be seen as a validation of these previous approaches in a more natural setting. However, the propagation distances can only represent a lower limit due to the inability of the shadowgraph method to visualize these cough airflows once their temperature cools to that of the ambient air, which is an important limitation of this methodology. The qualitative high-speed video footage of these volunteers coughing into their sleeves demonstrates that although this method rarely completely blocks the cough airflow, it decelerates, splits and redirects the airflow, eventually reducing its propagation. The effectiveness of this intervention depends on optimum positioning of the arm over the nose and mouth during coughing, though unsightly stains on sleeves may make it unacceptable to some.

Qualitative real-time schlieren and shadowgraph imaging of human exhaled airflows: an aid to aerosol infection control.

Using a newly constructed airflow imaging system, airflow patterns were visualized that were associated with common, everyday respiratory activities (e.g. breathing, talking, laughing, whistling). The effectiveness of various interventions (e.g. putting hands and tissues across the mouth and nose) to reduce the potential transmission of airborne infection, whilst coughing and sneezing, were also investigated. From the digital video footage recorded, it was seen that both coughing and sneezing are relatively poorly contained by commonly used configurations of single-handed shielding maneuvers. Only some but not all of the forward momentum of the cough and sneeze puffs are curtailed with various hand techniques, and the remaining momentum is disseminated in a large puff in the immediate vicinity of the cougher, which may still act as a nearby source of infection. The use of a tissue (in this case, 4-ply, opened and ready in the hand) proved to be surprisingly effective, though the effectiveness of this depends on the tissue remaining intact and not ripping apart. Interestingly, the use of a novel 'coughcatcher' device appears to be relatively effective in containing coughs and sneezes. One aspect that became evident during the experimental procedures was that the effectiveness of all of these barrier interventions is very much dependent on the speed with which the user can put them into position to cover the mouth and nose effectively.From these qualitative schlieren and shadowgraph imaging experiments, it is clear that making some effort to contain one's cough or sneeze puffs is worthwhile. Obviously, there will be a large amount of variation between individuals in the exact hand or tissue (the most common methods) configuration used for this and other practical factors may hinder such maneuvers in daily life, for example, when carrying shopping bags or managing young children.

Modeling Study of the Low-Temperature Oxidation of Large Methyl Esters from C11 to C19.

The modeling of the low temperature oxidation of large saturated methyl esters really representative of those found in biodiesel fuels has been investigated. Models have been developed for these species and then detailed kinetic mechanisms have been automatically generated using a new extended version of software EXGAS, which includes reactions specific to the chemistry of esters. A model generated for a binary mixture of n-decane and methyl palmitate was used to simulate experimental results obtained in a jet-stirred reactor for this fuel. This model predicts very well the reactivity of the fuel and the mole fraction profiles of most reaction products. This work also shows that a model for a middle size methyl ester such as methyl decanoate predicts fairly well the reactivity and the mole fractions of most species with a substantial decrease in computational time. Large n-alkanes such as n-hexadecane are also good surrogates for reproducing the reactivity of methyl esters, with an important gain in computational time, but they cannot account for the formation of specific products such as unsaturated esters or cyclic ethers with an ester function.