Diffusion of liquids into and through polymers is an important factor that negatively may influence the durability or lifetime of a polymer structure used in industrial applications. In this work two types of polymers, a liquid crystalline polymer (LCP, Vectra A950) and various thermoset resins as used in fibre reinforced plastics (FRP) process equipment were studied with regard to barrier properties, chemical resistance and long-term performance. LCP are known for their outstanding chemical resistance and barrier properties. FRP used in process equipment may be a cost-efficient solution in chemically aggressive environments where standard carbon or stainless steel cannot be used due to its limited corrosion resistance. Transport properties of typical industrial environments were determined for the LCP and the influence of annealing and orientation was investigated to study whether the barrier properties can be improved. The possibility to use LCP as cost-effective lining for FRP was explored. Special focus was put on the diffusion of water and its effect on long-term transport properties and stability of thermoset resins as the performance of FRP is strongly related to the diffusion of water. The results showed that Vectra A950 was suitable for organic solvent and non-oxidising acid environments. Its transport properties were gravimetrically determined and found to be 10 to 102 times lower than that of a high barrier fluoropolymer of type FEP. The degree of molecular packing increased with annealing time both below and above the melting point. Below the melting point this was – at least – partly due to crystal formation whereas above the melting point other mechanisms were involved. The effects of annealing and orientation on the transport properties in LCP were, however, very small or not significant and probably significantly longer annealing times are required. LCP has potential to be used as lining material for FRP as the use of an LCP-lining substantially reduced the permeability of and the solute sorption in a bisphenol A epoxy-based vinyl ester resin. The bonding strength was improved significantly by a combined abrasive and oxygen plasma treatment. The long-term sorption of water in thermoset resins including bisphenol A epoxy-based vinyl ester, novolac-based vinyl ester, urethane modified vinyl ester and bisphenol A polyester resins was found to increase with exposure time whereas the diffusion coefficient was not significantly affected. It was shown that the presence of water induced relaxation processes that were considered to be the primarily reason for the increase in sorption coefficient in comparison to degradation processes, such as hydrolysis, causing osmotic processes. A general relationship for the estimation of the sorption coefficient at 80ºC in dependence of the water activity and the sorption coefficient at unit activity independent of the resin type was established.

In this dissertation, we focus on two research topics related to mixture models. The first topic is Adaptive Rejection Metropolis Simulated Annealing for Detecting Global Maximum Regions, and the second topic is Bayesian Model Selection for Nonlinear Mixed Effects Model. In the first topic, we consider a finite mixture model, which is used to fit the data from heterogeneous populations for many applications. An Expectation Maximization (EM) algorithm and Markov Chain Monte Carlo (MCMC) are two popular methods to estimate parameters in a finite mixture model. However, both of the methods may converge to local maximum regions rather than the global maximum when multiple local maxima exist. In this dissertation, we propose a new approach, Adaptive Rejection Metropolis Simulated Annealing (ARMS annealing), to improve the EM algorithm and MCMC methods. Combining simulated annealing (SA) and adaptive rejection metropolis sampling (ARMS), ARMS annealing generate a set of proper starting points which help to reach all possible modes. ARMS uses a piecewise linear envelope function for a proposal distribution. Under the SA framework, we start with a set of proposal distributions, which are constructed by ARMS, and this method finds a set of proper starting points, which help to detect separate modes. We refer to this approach as ARMS annealing. By combining together ARMS annealing with the EM algorithm and with the Bayesian approach, respectively, we have proposed two approaches: an EM ARMS annealing algorithm and a Bayesian ARMS annealing approach. EM ARMS annealing implement the EM algorithm by using a set of starting points proposed by ARMS annealing. ARMS annealing also helps MCMC approaches determine starting points. Both approaches capture the global maximum region and estimate the parameters accurately. An illustrative example uses a survey data on the number of charitable donations. The second topic is related to the nonlinear mixed effects model (NLME). Typically a parametric NLME model requires strong assumptions which make the model less flexible and often are not satisfied in real applications. To allow the NLME model to have more flexible assumptions, we present three semiparametric Bayesian NLME models, constructed with Dirichlet process (DP) priors. Dirichlet process models often refer to an infinite mixture model. We propose a unified approach, the penalized posterior Bayes factor, for the purpose of model comparison. Using simulation studies, we compare the performance of two of the three semiparametric hierarchical Bayesian approaches with that of the parametric Bayesian approach. Simulation results suggest that our penalized posterior Bayes factor is a robust method for comparing hierarchical parametric and semiparametric models. An application to gastric emptying studies is used to demonstrate the advantage of our estimation and evaluation approaches. / Ph. D.

Modern machine learning methods are dependent on active optimization research to improve the set of methods available for the efficient and effective extraction of information from large datasets. This, in turn, requires an intense and rigorous study of optimization methods and their possible applications to crucial machine learning applications in order to advance the potential benefits of the field. This thesis provides a study of several modern optimization techniques and supplies a mathematical inquiry into the effectiveness of homotopy methods to attack a fundamental machine learning problem, effective clustering under constraints. The first part of this thesis provides an empirical survey of several popular optimization algorithms, along with one approach that is cutting-edge. These algorithms are tested against deeply challenging real-world problems with vast numbers of local minima, and compares and contrasts the benefits of each when confronted with problems of different natures. The second part of this thesis proposes a new homotopy map for use with constrained clustering problems. This thesis explores the connections between the map and the problem, providing several theorems to justify the use of the map and making use of modern homotopy tracking software to compare an optimization that employs the map with several modern approaches to solving the same problem. / Ph. D.

La Tesis Doctoral se fundamenta, principalmente, en la exploración de nuevos métodos de Control Predictivo Basado en Modelos (MBPC) mediante la incorporación de herramientas de optimización heurística y las mejoras en las prestaciones que se pueden conseguir con ello. La metodología de MBPC constituye un campo cada vez más importante en el control de procesos debido a que se trata de una formulación muy intuitiva, y a la vez muy potente, de un problema de control (por tanto es más fácilmente aceptable en el ámbito industrial). A pesar de ello, presenta limitaciones cuando se quiere aplicar a ciertos procesos complejos. Un elemento fundamental y al mismo tiempo limitante de ésta metodología lo constituye la técnica de optimización que se utilice. Simplificando mucho, el MBPC se convierte en un problema de minimización en cada periodo de muestreo, y la complejidad del problema de control se refleja directamente en la función a minimizar en cada instante. Si se incorporan modelos no lineales, restricciones en las variables, e índices de funcionamiento sofisticados, todo ello asociado a los problemas de tiempo real, se va a requerir algoritmos de optimización adecuados que garanticen el mínimo global en un tiempo acotado. En este sentido, la tesis incluye un análisis de las metodologías de Optimización Heurísticas, Simulated Annealing y Algoritmos Genéticos, como candidatas a la resolución de ese tipo de problemas y apartir de ellas realiza una implementación novedosa (denominada ASA) dentro del grupo de los algoritmos de Simulated Annealing que reduce el coste computacional. En los Algoritmos Genéticos, se obtienen las combinaciones de codificación y operadores genéticos más adecuadas para conseguir buenas relaciones de 'calidad de la solución/coste computacional' en la resolución de problemas de minimización complejos (no convexos, con discontinuidades, restricciones, etc.). Todo este análisis previo, permite la adaptación adecuada de estas técnicas heurísticas...... / Blasco Ferragud, FX. (1999). Control predictivo basado en modelos mediante técnicas de optimización heurística. Aplicación a procesos no lineales y multivariables [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/15995

[ES] Las maquinas industriales están fabricadas bajo unas especificaciones y tolerancias para cumplir sus funciones dentro del proceso productivo. Las maquinas producen vibraciones desde el inicio de su vida útil inherentes a las tolerancias e imperfecciones del proceso de fabricación de sus componentes, y que se incrementan por el desgaste de su uso. Las vibraciones provienen de las cargas dinámicas que producen las máquinas en su operación. Estas se transmiten a la cimentación y finalmente al suelo que lo soporta en forma de radiación. Un diseño de la cimentación y de los elementos de conexión entre la máquina y la cimentación (impedancias) reduce las vibraciones que transmite la máquina a su entorno. En el diseño de cimentaciones sometidas a fuerzas dinámicas hay muy pocas aportaciones dedicadas a la optimización, y no existen investigaciones sobre el transitorio de arranque de la máquina, momento en el que se producen las mayores vibraciones. El estado del arte concluye que los autores siempre han optimizado cimentaciones con un grado de libertad y en régimen de operación permanente. El cálculo dinámico de cimentaciones está orientado a reducir las vibraciones que se producen en el régimen permanente de operación. El diseño de las cimentaciones está basado en la aplicación de un conjunto de "Reglas de Oro" que conducen todas ellas a reducir las vibraciones aumentando la masa de la cimentación, es decir, su tamaño y en consecuencia su coste. En esta tesis se plantea un modelo de cimentación de 3 masas y 9 grados de libertad. Se determinan las ecuaciones del movimiento de las masas que componen el modelo a partir de la formulación de D'Alambert. Se integran en el tiempo empleando el método ß de Newmark. Esta metodología permite analizar además del régimen permanente de operación, el régimen transitorio de arranque de la máquina, limitando así el posible paso del conjunto por la zona de resonancia, no siendo necesario ya hacer el análisis modal en busca de la frecuencia de resonancia. El modelo contempla el movimiento vertical, el desplazamiento horizontal y el balanceo de la máquina, 3 grados de libertad para cada una de las masas que componen el modelo, considerando desplazamientos, velocidades y aceleraciones de cada una de las tres masas, habiendo por tanto 27 variables "solución" a analizar en cada caso. Una vez propuesto el modelo de cimentación, se han aplicado tres metaheurísticas diferentes para determinar la cimentación optima a partir de una función coste, sometida a las restricciones geométricas y las mecánicas que imponen las normas ISO 10816 y VDI 2056. El plan experimental se basa en la búsqueda aleatoria de soluciones factibles apli-cando métodos de vecindad, que cumplan las restricciones impuestas. El modelo contempla un vector solución de 37 variables que dan lugar a unos 8,12 1037 posibles configuraciones. Previo a la optimización metaheurística se estudia el espacio de soluciones apli-cando un proceso de Random Walk(paseo aleatorio) y obteniendo un conjunto de 2.000 posibles soluciones. La optimización se hace a partir de un mecanismo de búsqueda optima por entornos empleando tres metaheurísticas: Descent Local Search (DLS), Simulated Annealing (S.A.) y Late Acceptance Hill Climbing (L.A.H.C.). Se analiza comparativamente los resultados obtenidos, así como el comportamiento dinámico de las cimentaciones optimas calculadas. Como resultado se obtiene una reducción del coste de la cimentación de un 29% respecto al mejor valor obtenido aleatoriamente durante el Random Walk, que cumple las restricciones impuestas, tanto en régimen permanente de operación como en el régimen transitorio de arranque. Se concluye que las restricciones de vibración en el régimen transitorio de arranque garantizan el cumplimiento de las restricciones de vibración en el régimen permanente de operación. / [CA] Les maquines industrials estan fabricades sota unes especificacions i toleràncies per a complir les seues funcions dins del procés productiu. Les maquines produeixen vibracions des de l'inici de la seua vida útil inherents a les toleràncies i imperfeccions del procés de fabricació dels seus components, i que s'incrementen pel desgast del seu ús. Les vibracions provenen de les càrregues dinàmiques que produeixen les màquines en la seua operació. Aquestes es transmeten en forma de radiació a la fona-mentació i finalment al sòl que el suporta. Un disseny del model de fonamentació i dels elements de connexió entre la màquina i la fonamentació (impedàncies) redueix les vibracions que pateix la màquina i el seu entorn. En el disseny de fo-namentacions sotmeses a esforços dinàmics hi ha molt pocs investigació dedicada a l'optimització, i no existeixen investigacions sobre el transitori d'arrancada de la màquina, moment en el qual es produeixen les majors vibracions. L'estat de l'art conclou que els autors sempre han optimitzat fonamentacions amb un grau de llibertat i en règim d'operació permanent. El càlcul dinàmic de fonamentacions està orientat a reduir les vibracions que es produeixen en el règim permanent d'operació. El disseny de les fonamentacions està basat en l'aplicació d'una sèrie de "Regles d'Or" que condueixen totes elles a reduir les vibracions augmentant la massa de la fonamentació, és a dir, la seua grandària i en conseqüència el seu cost. En aquesta tesi es planteja un model de fonamentació de 3 masses. Es determinen les equacions de moviment de les masses a partir de la formulació de D'Alambert de la fonamentació. S'integren en el temps emprant el mètode ß de Newmark. Aquesta metodologia permet analitzar a més del règim permanent d'operació, el règim transitori d'arrancada de la màquina limitant així el possible pas del conjunt per la zona de ressonància, no sent per tant necessari ja fer l'anàlisi modal a la recerca de la freqüència de ressonància i el seu pas durant l'arrancada. El model contempla el moviment vertical, el desplaçament horitzontal y el balanç de la màquina. 3 graus de llibertat per cada una de les masses que compungeixen el model, considerant desplaçament, velocitat y acceleracions de cada una de las masses, finalment hi han 27 variables "solució" a analitzar cada vegada. Una vegada proposat el model de fonamentació, s'han aplicat tres metaheurísti-ques diferents per a determinar la fonamentació optima a partir d'una funció cost, somes a les restriccions geomètriques i mecàniques que imposan les normes ISO 10816 i VDI 2056. El pla experimental es basa en la cerca aleatòria de solucions factibles que com-plisca les restriccions imposades. El model contempla un total de 37 variables que donen lloc a uns 8,12 1037 possibles configuracions. Primerament, s'estudia l'espai de solucions mitjançant un procés de Random Walk analitzant un conjunt de 2.000 possibles solucions. L'optimització es fa a partir d'experiments de cerca optima per entorns emprant tres metaheurístiques: Descent Local Search (DLS), el Simulated Annealing (S.A.) i el Last Acceptance Hill Climbing (L.A.H.C.). S'analitza comparativament els resultats obtinguts, així com el comportament dinàmic de les fonamentacions optimes calculades. Com a resultat s'obté una reducció del cost de la fonamentació d'un 29% respecte al millor valor obtingut aleatòriament durant el Random Walk, que compleix tots els requeriments imposats, tant en règim permanent d'operació com en el règim transitori d'arrancada. Es conclou que les restriccions del regim transitori garan-teixen el compliment de les restriccions en règim permanent. / [EN] Industrial machines are made under specifications and tolerances to meet the requirements of the production process. Machines produces vibrations since the starting of its lifetime that are increased due to the wearing between the its parts. A suitable foundation reduces vibrations and its transmission to the environ. Vibrations are produced because of the dynamics loads produced by the ma-chines. These loads vary with time. These are transmitted as vibrations to the foundation and finally to ground. The design of the foundation and connection elements between the machines and the foundations (impedance) reduces the vibrations that the machines transmit to its environ. In the design of foundations under dynamic charges there are a few studies devoted to optimization, and there are not research on the transient starting, the moment when there are the major vibrations. The study of the art concludes that the authors always have optimized foundations with one degree of freedom and in permanent operation. The dynamic calculation of foundations always has been directed to reduce the vibrations in the permanent operation mode. The design of the foundations is under "rules of thumbs" that as results increases the mass of the foundation and its cost. In this thesis a model of 3 masses and 9 freedom degrees is proposed. The D'Alambert movement equations of the foundations are obtained and integrated using the Newmark's ß methodology. This methodology permits analysing the starting operation mode restricting its vibrations during the crossing of the reso-nance frequency. Model calculates the vertical movement, sliding and rocking, 3 degrees of free-dom for each one of the masses of the model, analysing displacements, speeds and accelerations of each one of the masses, in total 27 variable solution for each solution. Once the foundations model is proposed, optimization with three metaheuristics the foundation using a cost function, under mechanical and geometrical con-strains, proposed by the ISO 10816 y VDI 2056 Norms. The experimental plan is based in the searching of feasible solutions under the constrains. The foundation model depends on 37 variables, making 8,12 1037 pos-sible sets of configurations. Firstly, the space of solutions is analysed with the technique of Radom Walk. A set of 2000 solutions was calculated. For searching the optimal solution three metaheuristics are used: Descent Local Search (DLS), Simulated Annealing (SA) and Last Acceptance Hill Climbing methodologies were selected and the results are compared, and the dynamic behaviour of the foundations calculated. As a result, a reduction of the 29% of the cost of the foundation is obtained in relation with the best cost obtained randomly during the Random Walk that achieves the constrains applied in transient and permanent operation mode. It's concluded that the constrains of transient mode guarantees the compliance of the constrains in permanent operation mode. / Terrádez Marco, JL. (2024). Optimización de un modelo de cimentación de una máquina industrial rotativa [Tesis doctoral]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/211831

No description available.

No description available.

程序式動畫是一種根據使用者所提供的高階運動參數，自動產生動畫的方法。藉著高階的運動參數，程序式動畫比運動擷取資料有著更高的彈性。使用者可透過調整參數，輕易地讓動畫滿足情境上所需的限制。但如何調整適當的運動參數以產生擬真的動畫仍屬不易，因此程序式動畫常有在視覺上觀感不自然的問題。本研究的目標是，將運動擷取資料擬真的要素，帶到程序式動畫之中，以改進程序式動畫的品質。我們將問題定義成一個最佳化問題：給定一段運動擷取資料，系統該如何調整程序式動畫之參數，使得程序式動畫與運動擷取資料的差距盡可能地縮小？我們的系統可以參考一段運動擷取資料，以最佳化演算法，自動調整程序式動畫的參數，搜尋能產生出與運動擷取資料最為相似的運動參數。為了進一步讓產生之動畫符合環境的限制需求，多組最佳化過後的運動參數可以再透過內插，重新產生出一組符合限制需求的運動參數。實驗結果顯示，我們的方法不但使程序式動畫得以保留原來彈性的優點，也改善了程序式動畫常有的視覺觀感不自然的缺點。 / Procedural animation provides a way for a user to generate animation according to the high-level motion parameters that the user supplies. With the high-level motion parameters, procedural animation has the flexibility of generating animation accommodating the requested constraints in a scenario. However, tuning parameters to generate realistic animations usually is a difficult task. Therefore, animations produced with this approach often have the drawback of unrealistic-looking. Our goal is to improve the quality of procedural animation by bringing the naturalness of motion capture data into procedural animation. We model our problem as an optimization problem: given a motion captured clip, how does the system tune the motion parameters in an animation procedure to minimize the difference between animations produced by a procedure and captured in a motion clip? Our proposed system takes a motion captured clip as a reference and tunes the motion parameters of the animation procedure with an optimization algorithm. In order to generate animation satisfying environmental constraints, multiple optimized motion parameters can be interpolated to create a new set of motion parameters which can also satisfy the constraints. Our experimental results show that our method not only retains the flexibility of procedural animation, but also enhances the quality of procedural animation.

In this thesis, gas sensors based on on-chip hydrothermally grown 1-D zinc oxide (ZnO) nanostructures are presented, to improve the sensitivity, selectivity, and stability of the gas sensors. Metal-oxide-semiconductor (MOS) gas sensors are well-established tools for the monitoring of air quality indoors and outdoors. In recent years, the use of 1-D metal oxide nanostructures for sensing toxic gases, such as nitrogen dioxide, ammonia, and hydrogen, has gained significant attention. However, low-dimensional nanorod (NR) gas sensors can be enhanced further. Most works synthesize the NRs first and then transfer them onto electrodes to produce gas sensors, thereby resulting in large batch-to-batch difference. Therefore, in this thesis six studies on 1-D ZnO NR gas sensors were carried out. First, ultrathin secondary ZnO nanowires (NWs) were successfully grown on a silicon substrate. Second, an on-chip hydrothermally grown ZnO NR gas sensor was developed on a glass substrate. Its performance with regard to sensing nitrogen dioxide and three reductive gases, namely, ethanol, hydrogen, and ammonia, was tested. Third, three 1-D ZnO nanostructures, namely, ZnO NRs, dense ZnO NWs, and sparse ZnO NWs, were synthesized and tested toward nitrogen dioxide. Fourth, hydrothermally grown ZnO NRs, chemical vapor deposited ZnO NWs, and thermal deposited ZnO nanoparticles (NPs) were tested toward ethanol. Fifth, the effect of annealing on the sensitivity and stability of ZnO NR gas sensors was examined. Sixth, ZnO NRs were decorated with palladium oxide NPs and tested toward hydrogen at high temperature. The following conclusions can be drawn from the work in this thesis: 1) ZnO NWs can be obtained by using a precursor at low concentration, temperature of 90 °C, and long reaction time. 2) ZnO NR gas sensors have better selectivity to nitrogen dioxide compared with ethanol, ammonia, and hydrogen. 3) Sparse ZnO NWs are highly sensitive to nitrogen dioxide compared with dense ZnO NWs and ZnO NRs. 4) ZnO NPs have the highest sensitivity to ethanol compared with dense ZnO NWs and ZnO NRs. The sensitivity of the NPs is due to their small grain sizes and large surface areas. 5) ZnO NRs annealed at 600 °C have lower sensitivity toward nitrogen dioxide but higher long-term stability compared with those annealed at 400 °C. 6) When decorated with palladium oxide, both materials form alloy at a temperature higher than 350 °C and decrease the amount of ZnO, which is the sensing material toward hydrogen. Thus, controlling the amount of palladium oxide on ZnO NRs is necessary.

La synthèse de siliciures métalliques sous la forme de films ultra-minces demeure un enjeu majeur en technologie CMOS. Le contrôle du budget thermique, afin de limiter la diffusion des dopants, est essentiel. Des techniques de recuit ultra-rapide sont alors couramment utilisées. Dans ce contexte, la technique de nanocalorimétrie est employée afin d'étudier, in situ, la formation en phase solide des siliciures de Ni à des taux de chauffage aussi élevés que 10^5 K/s. Des films de Ni, compris entre 9.3 et 0.3 nm sont déposés sur des calorimètres avec un substrat de a-Si ou de Si(100). Des mesures de diffraction de rayons X, balayées en température à 3 K/s, permettent de comparer les séquences de phase obtenues à bas taux de chauffage sur des échantillons de contrôle et à ultra-haut taux de chauffage sur les calorimètres. En premier lieu, il est apparu que l'emploi de calorimètres de type c-NC, munis d'une couche de 340 nm de Si(100), présente un défi majeur : un signal endothermique anormal vient fausser la mesure à haute température. Des micro-défauts au sein de la membrane de SiNx créent des courts-circuits entre la bande chauffante de Pt du calorimètre et l'échantillon métallique. Ce phénomène diminue avec l'épaisseur de l'échantillon et n'a pas d'effet en dessous de 400 °C tant que les porteurs de charge intrinsèques au Si ne sont pas activés. Il est possible de corriger la mesure de taux de chaleur en fonction de la température avec une incertitude de 12 °C. En ce qui a trait à la formation des siliciures de Ni à ultra-haut taux de chauffage, l'étude montre que la séquence de phase est modifiée. Les phases riches en m étal, Ni2Si et théta, ne sont pas détectées sur Si(100) et la cinétique de formation favorise une amorphisation en phase solide en début de réaction. Les enthalpies de formation pour les couches de Ni inférieures à 10 nm sont globalement plus élevées que dans le cas volumique, jusqu' à 66 %. De plus, les mesures calorimétriques montrent clairement un signal endothermique à haute température, témoignant de la compétition que se livrent la réaction de phase et l'agglomération de la couche. Pour les échantillons recuits a 3 K/s sur Si(100), une épaisseur critique telle que décrite par Zhang et Luo, et proche de 4 nm de Ni, est supposée. Un modèle est proposé, basé sur la difficulté de diffusion des composants entre des grains de plus en plus petits, afin d'expliquer la stabilité accrue des couches de plus en plus fines. Cette stabilité est également observée par nanocalorimétrie à travers le signal endothermique. Ce dernier se décale vers les hautes températures quand l'épaisseur du film diminue. En outre, une 2e épaisseur critique, d'environ 1 nm de Ni, est remarquée. En dessous, une seule phase semble se former au-dessus de 400 °C, supposément du NiSi2. / The synthesis of metal silicides in the form of ultra-thin films remains a major issue in CMOS technology. Controlling the thermal budget to limit dopant diffusion is essential and ultra-rapid annealing techniques are commonly used. In this context, the nanocalorimetry technique is used in situ to study Ni silicides phase formation at heating rate as high as 10^5 K/s. Ni films from 9.3 nm to 0.3 nm are carried out on calorimeters with either a-Si or Si(100) substrate. Ex situ temperature-resolved x-ray diffraction, scanned at low heating rate (3 K/s), is used in order to compare the phase sequence with the ones obtained at ultra-high heating rates. First, the use of c-NC calorimeters equipped with a Si(100) substrate, 340 nm thick, results in a major challenge. An abnormal endothermic signal is measured at high temperature. Results show that microdefects in the SiNx membrane create short circuits between the Pt heating strip of the calorimeter and the metal sample. This phenomenon decreases with the thickness of the metal sample and has no effect below 400 °C. Below this temperature the Si charge carriers are not activated. Then, the heat flow signal was corrected as a function of the temperature with an uncertainty of 12 ° C. In regard to the formation of Ni silicides at ultra-high heating rates, the study shows that the phase sequence is changed. On Si(100) substrate, the metal rich phases such as Ni2Si and theta -phase are no longer detected. Also, a solid state amorphization is favored at the beginning of the reaction. The enthalpies of formation for layers thinner than 10 nm are generally higher, up to 66 %, than those listed in literature for thick layers. Moreover, the calorimetric measurements clearly show an endothermic signal at high temperature, revealing a competition between the reaction of phase formation and the layer agglomeration. For sample annealed at 3 K/s on Si(100), the critical thickness as described by Zhang and Luo, is observed around 4 nm of Ni. A model is proposed in order to explain the increased stability when the layer thicknesses decrease. It is based on the fact that the diffusion of atoms is harder when grain sizes decrease. It also explains why the endothermic signals in nanocalorimetry shift toward higher temperatures when the layer thickness decreases. In addition, a second critical thickness of about 1 nm is noted. Below, it seems that only one phase forms, presumably the NiSi2.