VTT Research System
Not a member yet
161399 research outputs found
Sort by
Efficient debromination of tetrabromobisphenol A in protic solvents by supported nickel catalysts:Effect of metal-support interactions
No full textToxic and environmentally hazardous brominated flame retardants (BFR) hinder the recycling of plastic waste, which has led to the development of various extraction processes to remove them. These processes can be further advanced by debrominating BFRs into less harmful compounds with potential commercial value, thus supporting the principles of a circular economy. Herein, Ni/Al2O3 was employed for the catalytic debromination of tetrabromobisphenol A flame retardant in mixtures of H2O, isopropanol and NaOH at modest reaction temperatures. Activity experiments conducted in an autoclave indicated that studied catalyst exhibits impressive debromination activity and complete selectivity towards C−Br bond scission. The catalyst reduction temperature was found to correlate with debromination activity, with higher temperatures yielding improved performance. Debromination proceeded under H2 and also under N2 in protic solvents via transfer hydrogenation. Catalyst characterization, coupled with high-resolution mass-spectrometry product analytics and deuterium labelling, suggested that the enhanced catalytic activity can be attributed to the activation of the metal-support interface and subsequent interactions with adsorbed solvent molecules and associated dissociation products on the alumina support. Used experimental conditions also provided high tolerance against bromine poisoning of the catalyst, in contrast to reference debromination experiments conducted in toluene. The study demonstrates the capability of Ni/Al2O3 as an efficient and affordable debromination catalyst in solvents of low environmental impact. Furthermore, the results provide additional insights into structure-activity relationships of supported nickel catalysts in protic solvents, which can be leveraged for the development of more efficient and sustainable dehalogenation and heteroatom removal processes for environmental applications.</p
Emulating a forest growth and productivity model with deep learning
No full textWe studied the possibility of replacing a complex forest growth and productivity model with a deep learning model with sufficient accuracy. We used three different neural network architectures for emulating the prediction task of the PREBASSO (Mäkelä 1997; Minunno et al. 2016) forest growth model: 1) Recurrent Neural Network (RNN) Encoder-decoder network, 2) RNN encoder network, and 3) Transformer encoder network. The PREBASSO forest growth model was used to produce 25-year predictions for forest variables: tree height, stem diameter, basal area, and the carbon balance variables: net primary production (NPP), gross primary production per tree layer (GPP), net ecosystem exchange (NEE) and gross growth (GGR) to train the machine learning models. The Finnish Forest Centre provided the data for 29 619 field inventory plots in continental Finland that were used as the initial state of the forest sites to be simulated. Climate data downloaded from Copernicus Climate Data Store were used to provide realistic climate scenarios. We emphasized the importance of low bias in long term predictions and set the goal for the emulator prediction relative bias to be within ±2%. The RNN encoder model produced the best results with the mean of the yearly bias values within the specified ±2% limit over the 25-year prediction period. The study shows that emulating the operation of analytical forest growth models is feasible using state-of-the-art machine learning methods and indicates the potential of using such emulators for producing long time span simulations for e.g. digital twins.</p
Advanced Assessment of Stroke in Retinal Fundus Imaging with Deep Multi-view Learning
No full textStroke is globally a major cause of mortality and morbidity, and hence, accurate risk assessment and diagnosis of stroke are valuable. Retinal fundus imaging reveals the known markers of elevated stroke risk in the eyes, which are retinal venular widening, arteriolar narrowing, and increased tortuosity. In contrast to other imaging techniques used for stroke assessment, the acquisition of fundus images is easy, non-invasive, fast, and inexpensive. This paper examines the feasibility of utilizing retinal fundus imaging to differentiate individuals with stroke or transient ischemic attack (TIA), aiming to assess its potential for screening or diagnostic applications. Therefore, in this study, we propose a multi-view stroke network (MVS-Net) to detect stroke and TIA using retinal fundus images. Contrary to existing studies, our study proposes for the first time a solution to discriminate stroke and TIA with deep multi-view learning by proposing an end-to-end deep network, consisting of multi-view inputs of fundus images captured from both right and left eyes. Accordingly, the proposed MVS-Net defines representative features from fundus images of both eyes and determines the relation within their macula-centered and optic nerve head-centered views. Experiments performed on a dataset collected from stroke and TIA patients, in addition to healthy controls, show that the proposed framework achieves an AUC score of 0.84 for stroke and TIA detection.</p
Mineral carbonation:Thermal activation of serpentine as flexible component in energy systems
No full textIn this article, the thermal activation of serpentine for mineral carbonation using an electrically heated fluidised bed was studied. This process step is combined into a novel process configuration with rapid thermal activation and a thermal energy storage. For the studied serpentinite mine tailings, the thermal activation durations corresponding with desired mineralogical changes were ' 2 min at 700 °C, 4–8 min at 650 °C, 16–75 min at 600 °C and ' 75 min at 550°C, improving reaction kinetics compared to chamber furnace processing at 650°C for 30 min or 750 °C for 15 min. In all fluidised bed experiments, the preferred mineralogical changes corresponded to a dehydroxylation degree of 52–72 % determined with simultaneous thermal analysis. Fourier-transform infrared analyser was used for online determination of the dehydroxylation degree and could be used for process control if the measurement delay was considered better. The mass and energy balances for an industrial-scale thermal activation plant matching 100,000 tCO2/a storage were calculated. The average electrical power demand was 17.4 MW, of which 22.9 % was recovered from steam and 61.7 % was stored in the thermal energy storage silos. Twelve-hour and 7-day buffering times resulted in silo volumes of 452.5 m3 and 3167 m3 and thermal energy storage capacities of 129.0 MWh and 903.2 MWh, respectively. Further studies should include process parameter optimisation as well as techno-economic and life cycle assessments to holistically improve the concept.</p
Penalty force stabilization method for elasto-plastic correspondence models in peridynamics
No full textZero-energy modes are a challenge in non-ordinary state-based peridynamics. This work extends a penalty force approach to stabilize such models in both elastic and elasto-plastic simulations. The method penalizes nonuniform deformation directly through a tangent-modulus-based correction that adapts to the evolving material state. To evaluate its performance, we introduce two novel zero-energy mode measures — nodal and global nonuniform strain. We compare the method with first- and second-order bond-associated formulations in small and finite strain regimes and assess the influence of power-law, Gaussian, and uniform weight functions. In small strain tests, the penalty force method matches analytical solutions with displacement errors below 10 −9 with negligible zero-energy mode measures. In finite strain plasticity, the method converges reliably, reproduces finite element and experimental stress–strain responses, and maintains global displacement errors below 10 −4. It shows low sensitivity to the choice of a peridynamic weight function. The penalty force method requires only one deformation gradient evaluation per node, avoids tuning parameters, and suppresses zero-energy artifacts to negligible levels. The results show that it provides a stable and efficient alternative for correspondence-based peridynamic simulations across a wide range of deformation regimes.</p
Emulating a forest growth and productivity model with deep learning
No full textWe studied the possibility of replacing a complex forest growth and productivity model with a deep learning model with sufficient accuracy. We used three different neural network architectures for emulating the prediction task of the PREBASSO (Mäkelä 1997; Minunno et al. 2016) forest growth model: 1) Recurrent Neural Network (RNN) Encoder-decoder network, 2) RNN encoder network, and 3) Transformer encoder network. The PREBASSO forest growth model was used to produce 25-year predictions for forest variables: tree height, stem diameter, basal area, and the carbon balance variables: net primary production (NPP), gross primary production per tree layer (GPP), net ecosystem exchange (NEE) and gross growth (GGR) to train the machine learning models. The Finnish Forest Centre provided the data for 29 619 field inventory plots in continental Finland that were used as the initial state of the forest sites to be simulated. Climate data downloaded from Copernicus Climate Data Store were used to provide realistic climate scenarios. We emphasized the importance of low bias in long term predictions and set the goal for the emulator prediction relative bias to be within ±2%. The RNN encoder model produced the best results with the mean of the yearly bias values within the specified ±2% limit over the 25-year prediction period. The study shows that emulating the operation of analytical forest growth models is feasible using state-of-the-art machine learning methods and indicates the potential of using such emulators for producing long time span simulations for e.g. digital twins.</p
Modeling Thermal Effects in Atomic Layer Deposition for Trench-Shaped Structures
No full textAn atomic layer deposition (ALD) simulation approach is presented for transient diffusion of heat and mass at low Knudsen numbers (Kn < 0.1), focusing on thermal effects in trench-shaped structures. Two boundary conditions (BCs) are analyzed: the ‘thin wall’ BC incorporates exothermic reactions with a derived wall heat flux term, and the ‘thick wall’ BC maintains constant wall temperature ranging between 500 K and 800 K. For both BCs, we examine aspect ratios from 1 to 100. The chosen BC significantly impacts reaction kinetics/peak temperatures, with local temperature variations up to 200 K under ‘thin wall’ conditions. The coating time ratio between ‘thin wall’ and ‘thick wall’ ranges from 0.9 to 1.7. Two ‘universal’ functional forms are proposed to explain how surface coverage depends on time and how coating time relates to aspect ratio and diffusion timescale. Results emphasize the crucial role of temperature distribution in ALD, impacting growth per cycle, reactant decomposition/desorption, and potential substrate damage
Hydroxylamine grafting of periodate oxidized cellulose microfibrils and its impact on fibre adhesion
No full textIn this work, dialdehyde cellulose microfibrils (DA-CMFs) were reacted with O-substituted hydroxylamines, demonstrating an effective and versatile method for lateral functionalization of DA-CMFs using mild, aqueous reaction conditions. Depending on the conditions used, partial or complete substitution of the aldehydes could be achieved. The reaction was performed in the presence and absence of the reducing agent α-picoline borane (PB). DA-CMFs were reacted with O-(carboxymethyl) hydroxylamine (HAAA), and the adhesive properties of native and HAAA-conjugated DA-CMFs to fibres were studied. The adhesive properties were shown to depend on charge; while native DA-CMFs aggregate heavily in contact with the fibre surface, HAAA-conjugated DA-CMFs showed significantly improved adhesion. For a degree of substitution of 50% or higher, a sealed layer, without aggregates, could be observed using electron microscopy. Finally, a versatile protocol for co-insertion of HAAA (carboxylate) and aminooxy-PEG3-azide (hydroxylamine azide) was developed and demonstrated, with high yields of insertion. The modified DA-CMFs retained good adhesive properties to fibres, showing that the approach is general, and that the chemistry can be tuned depending on the target application
Quasioptic, Calibrated, Full 2-port Measurements of Cryogenic Devices under Vacuum in the 220- 330 GHz Band
No full textA quasi-optical (QO) test bench was designed, simulated, and calibrated for characterizing all four S-parameters of devices in the 220–330 GHz (WR3.4) frequency range, from room temperature down to 4.8 K. Quasioptical calibration methods were applied to de-embed the impact of cryostat and optical elements on device under test measurements. The devices were measured through vacuum windows via focused beam radiation. A de-embedding method employing line-reflect-match (LRM) calibration was established to account for the effects of optical components and vacuum windows. Such a method does not require multiple line standards inside the cryostat and mechanical translation of quasioptics. System validation was performed with measurements of cryogenically cooled devices, such as bare silicon wafers and stainless-steel frequency-selective surface (FSS) bandpass filters, and superconducting bandpass FSS fabricated in niobium. A permittivity reduction of Si based on a 4 GHz resonance shift was observed concomitant with a drop in temperature from 296 to 4.8 K. The stainless steel FSS measurements revealed a relatively temperature invariant center frequency and return loss level of 263 GHz and 35 dB on average, respectively. Finally, a center frequency of 257 GHz was measured with the superconducting filters, with return loss improved by 11 dB on average at 4.8 K. To the best of our knowledge, this is the first reported attempt to scale LRM calibration to 330 GHz and use it to de-embed the impact of optics and cryostat from cryogenically cooled device S-parameters.</p
Cascade synthesis of diarylamines catalyzed by oxygen-rich and porous carbon
No full textActivated carbon derived porous materials, effectively enriched with OH and C[double bond, length as m-dash]O groups, were found to mediate, in a cascade manner, the condensation between anilines and 3-hexenones or β-tetralones, followed by their aromatization to diarylamines. The reaction proceeds via in situ formation of enamine intermediates which are subsequently oxidatively dehydrogenated in presence of a molecular oxidant under inert atmosphere. The functional groups on the carbon surface contributed actively to the catalysis: phenolic hydroxyl groups were found to promote the coupling of amines and ketones to imines and their tautomerization to enamines, while the C[double bond, length as m-dash]O groups of the quinoidic moieties catalyze the dehydrogenative aromatization step. The carbon material's extensive porous structure turns out to be critical to preserve the reactive β,γ-unsaturated cyclohexanone derivatives and their enamine intermediates from undesirable coupling and condensation side-reactions. The carbocatalyst can be regenerated by molecular N-oxo quinoline, which acts as a more convenient and cleaner stoichiometric oxidant in comparison with standard aerobic conditions (oxygen-rich atmosphere). The developed methodology delivered up to 93% yields for many diarylamines, formerly accessible exclusively via Pd-mediated couplings. Computational DFT study of possible enamine reaction modes with quinone model compounds, combined with kinetic isotope effects (KIE) suggest that the aromatization reaction is triggered by hydride abstraction at the benzylic position of the enamine intermediate