Two Short Term Applications of the Semantic Web — Working Paper (Draft)

合集下载

Study of the kinetics of nucleation and growth

Study of the kinetics of nucleationand growthNucleation and growth are two fundamental processes that occur in many systems, from chemical reactions to the formation of crystals. Understanding the kinetics of nucleation and growth is essential for predicting and controlling the properties of materials. In this article, we will explore the principles behind these processes and examine some of the experimental techniques used to study them.First, let us consider nucleation, which is the process by which a new phase or crystal is formed from a homogeneous solution or gas. Nucleation occurs when the concentration of particles in the solution or gas exceeds a certain critical value, leading to the formation of small clusters of particles. These clusters continue to grow by the addition of more particles until they reach a critical size, at which point they are stable and can continue to grow without further nucleation.The kinetics of nucleation can be described using a variety of models, depending on the nature of the system and the experimental conditions. One commonly used model is the classical nucleation theory, which assumes that the nucleation rate is proportional to the concentration of particles in the solution and the free energy barrier for nucleation. The free energy barrier depends on factors such as the surface energy between the new phase and the existing phase, the size of the critical nucleus, and the temperature of the system.Experimental methods for studying nucleation include microscopy techniques such as transmission electron microscopy (TEM) and scanning electron microscopy (SEM), which can be used to observe the formation and growth of clusters. X-ray diffraction (XRD) can also be used to identify the crystal structure of the new phase.Now let us turn to growth, which is the process by which the clusters formed during nucleation continue to increase in size. Growth occurs when particles in the solution or gas are able to attach to the surface of the cluster and become incorporated into thecrystal lattice. The rate of crystal growth is determined by the concentration of particles in the solution or gas, the surface area of the crystal, and the diffusion coefficient of the particles.The kinetics of growth can be described using the Lifshitz-Slyozov-Wagner (LSW) theory, which assumes that the rate of crystal growth is inversely proportional to the cube of the particle size. This means that smaller particles grow faster than larger particles, leading to a decrease in the overall particle size distribution over time.Experimental methods for studying crystal growth include techniques such as time-resolved XRD, which can be used to monitor the evolution of the crystal structure over time. In situ optical microscopy can also be used to observe the growth of individual crystals in real time.In summary, the study of the kinetics of nucleation and growth is essential for understanding the behavior of materials in a wide range of applications. The principles behind these processes are complex, but can be described using mathematical models such as classical nucleation theory and the LSW theory. A variety of experimental techniques are available to study these processes, including microscopy, XRD, and optical techniques. By combining theoretical models with experimental data, researchers can gain a detailed understanding of the mechanisms behind nucleation and growth, and develop new materials with tailored properties.。

Materials Characterization

Materials Characterization Materials characterization is a crucial aspect of material science and engineering. It involves the study of the physical and chemical properties of materials and their behavior under different conditions. The characterization of materials is essential for the development of new materials and the improvement of existing ones. In this essay, I will discuss the importance of materials characterization, the different techniques used for materials characterization, and the challenges faced in materials characterization.Materials characterization is important because it provides information about the structure and properties of materials. This information is necessary for the development of new materials that can meet specific requirements. For example, materials used in aerospace applications must be lightweight, strong, andresistant to high temperatures. Materials characterization helps researchers understand the properties of different materials and how they can be modified to meet these requirements. It also helps in the identification of defects and impurities in materials, which can affect their performance.There are several techniques used for materials characterization, each withits advantages and limitations. One of the most commonly used techniques is microscopy, which allows researchers to observe the structure of materials at the micro and nanoscale. Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) are two types of microscopy used for materials characterization. SEM provides high-resolution images of the surface of materials, while TEM provides images of the internal structure of materials.Another technique used for materials characterization is X-ray diffraction (XRD), which provides information about the crystal structure of materials. XRD is used to identify the phases present in a material, as well as their orientation and lattice parameters. X-ray fluorescence (XRF) is another technique used for materials characterization, which provides information about the elemental composition of materials. XRF is used to identify impurities and contaminants in materials.Despite the importance of materials characterization, there are several challenges faced in this field. One of the challenges is the complexity ofmaterials. Many materials have complex structures, making it difficult to obtain accurate information about their properties. Another challenge is the need for specialized equipment and expertise. Materials characterization requires expensive equipment and highly skilled personnel, which can be a barrier to entry for researchers.In addition, materials characterization can be time-consuming and expensive.It may take several weeks or even months to obtain accurate data about a material. This can delay the development of new materials and increase the cost of research. Furthermore, materials characterization may require destructive testing, which can limit the amount of material available for further analysis.In conclusion, materials characterization is a crucial aspect of material science and engineering. It provides information about the properties of materials and their behavior under different conditions. There are several techniques usedfor materials characterization, each with its advantages and limitations. However, there are also challenges faced in this field, such as the complexity of materials, the need for specialized equipment and expertise, and the time and cost involvedin obtaining accurate data. Despite these challenges, materials characterization remains an essential tool for the development of new materials and the improvement of existing ones.。

Electron Optic Design of Arrayed E-Beam Microcolumns Based Systems for Wafer Defects Inspec

a r X i v :0805.0495v 1 [p h y s i c s .o p t i c s ] 5 M a y 2008ELECTRON OPTIC DESIGN OF ARRAYEDE-BEAM MICROCOLUMNS BASED SYSTEMS FORW AFER DEFECTS INSPECTIONV.V.Kazmiruk,T.N.SavitskajaInstitute of Microelectronics Technology and High Purity Materials,Russian Academyof SciencesAbstractIn this paper is considered a matter of the system for wafer defect inspection (WDIS)practical realization.Such systems are on the agenda as the next generation and substitution for light optics and single e -beam based WDISs.Introduction At the present time an activity in the ﬁeld of e-beam micrcolumns practical realization is growing up rapidly.The most signiﬁcant progress is attained by groups of T.H.P.Chang from IBM Research Center [1],P.Kruit from Delft Technical University [2–6]and H.S.Kim and alii [7,8].However,their eﬀorts directed mainly on e -beam lithography application or just microcolumn electron optics design.In this paper is considered a matter of the system for wafer defect inspection (WDIS)practical realization.Such systems are on the agenda as the next generation and substitution for light optics and single e -beam based WDISs.In our previous work [9]the requirements to WDIS have been considered as informative system with resolution down to 2nm.It was shown that in the range of 10÷30nm multibeam WDIS for topographical defects inspection would be comparable in throughput with the light optics system when number of columns in the array is about 1000.In the case of the line width mea-suring (LWM)or surface microrelief reconstruction can be realized resolution 2–10nm.Are considered aspects of WDIS design for both application.The electron optics design First of all consider the main principles of electron optics design of the microcolumn.We start from simple single lens column used by many authors[1,8]for experiments in this ﬁeld.The electron optical components of a one lens column are shown schemat-ically in ﬁg.1.The resolution of the microscope column is limited primarily by the aber-rations of the objective lens.The probe size is given by:d 2b =(M ·d 0)2+d 2d +d 2s +d 2c ,(1)where M is the column magniﬁcation;d 0is the virtual source size;d d =1,5·αV (0,5)(2)Fig.1.e-beam lithographyis the diameter of the diﬀraction disk so that d d/2is full width half maximum of corresponding distribution.d s=1/2·C sα3is the spherical aberration disc with the spherical aberration coeﬃcient C s;d c=C cα∆V/Vis the chromatic aberration disk with C c being aberration coeﬃcient and DV being the energy spread of the beam;α0=α·M,(3) whereα0is the semi convergent angle at the exit of the source andαis the semi convergent angle at the target.Final probe current I b isI b=pα20dI0/dΩ0.(4) Here dI0/dΩ0is angular emission density.We use the conventional rule(1)to estimate the chromatic and spherical aberration coeﬃcients of the objective lens:C c and C s.Fig.2shows the performance of1keV microcolumns with two diﬀerent objective lenses[1].Theﬁrst in solid line,represents aﬁxed symmetric einzel lens.This lens has a200µm bore diameter and250µm spacing.The lens. operating for a1mm working distance in the accelerating mode,has a chro-matic and spherical abeiration coeﬄcients of approximately2mm and50mm respectively.As shown in theﬁgure,a probe size of9,9nm can be achieved al an optimum semiconvergent angle of≈6,3mrad.Further improvement of resolution can be achieved by optimizing the electrodes geometry for work-ing distance1mm,that allows to decrease both spherical and chromatical coeﬃcients to values shown by dashed lines.As a result the resolution8,8nmFig.2.at working distance1mm can be achieved.These are typical results achieved practically so far[1,8].It should be noted that those results achieved in transmission mode,and working distance1mm is chosen to place on-axis detector between lens and sample as it shown inﬁg.3.Fig.with on-axis detector Now consider what should be changed for improving a resolution to2nm.Is assumed that electrons energy still is1keV and the energy spread∆V=1eV. In theﬁg.4is shown an electron optical performance of1keV improved column for C s=0,3mm,C c=0,084mm.101010-1012a -image (radians)2,4 10Fig.4.Electron optical performance of 1keV improved column.C s =0,3mm,C c =0,084mmIt is obvious from(1)that the value d b =2nm can be achieved when each of d d ,d s ,d c ,is less than that value.Thus,diﬀraction limit becomes a dominating factor which determines semi convergent angle.If is chosen α≫2,4·10−2radians,then aberration coeﬃcients C s ≪0,3mm and C c ≪0,08mm.For more exact evaluation examine the residuald 2−(d 2d +d 2c +d 2s )=(M d 0)2at d b =2nm and minimize (d 2d +d 2c +d 2s )over α.Thus for given aberration coeﬃcients C s and C c the maximum value of (M d 0)2can be calculated.Figures 5,6show result for C c =0,04and C c =0,02mm.x 104C s nmnm 2Fig.5.nm2Fig.6.Thus,to receive the probe size2nm for objective lens with the chromatic aberration coeﬃcient C c=0,04,the spherical aberration coeﬃcient C s needs to be lower0,02mm and semiconvergent angleα>2,6·10−2rad.Similarly for C c=0,02mm C s≤0,08mm andα>2,7·10−2rad is required.The use another formulas for probe size diameter calculation,for example [2],gives no principal change to order of C s and C c values.Such a way from the above analytical performance consideration follows that for improvement of the resolution to2nm it’s necessary to keep semi con-vergent angle more than2,7·10−2rad and radically decrease both chromatical and spherical aberrations coeﬃcients.Methods of improvement aberrations consist in electrostatic lens dimen-sional scaling down from conventional lens.Inﬁg.7is schematically shown spherical aberration for a positive and a negative lens,illustrated with two rays entering the lens at diﬀerent radii,r1and r2.In both cases,the inter-cept with the z-axis shifts in the negative z direction for increasing radius incidence.ofAs working distance WD=r/α,then for incidence radius in the range2÷10µm WD should be in the range74÷370µm.Such short working distance does not give enough space for detector placement.In practiceαis even more than2,7·10−2rad in order to obtain small diﬀraction term.Table1.Two lens system performanceProbe sizeProbe currentMagniﬁcation(specimen–gun)C s gun sideC c gun sideSpherical aberration termChromatic aberration termDiﬀraction termUltra thinﬁlm foil implementation for improvement the miniature beam system performanceThere are two promising applications of ultra thin foils for electron mi-croscopy:the tunnel junction emitter and the low energy mon for both applications is that the electron beam is sent through the thin foil at low energy.Measurements of mean free path for number of metals indicate the value about5nm at the energy≈5eV above the Fermi level.First achieved by us[6]free standing foils have been5nm of thickness and later we achieved foils with thickness4,3and even2,2nm.A substantial part of electrons can be transmitted through such thinﬁlm without scattering,so ﬁlm acts as an ideal energyﬁlter.The tunnel junction emitterElectronﬁeld emitters are used in a wide variety of applications,such as:electron microscopes,electron beam lithography machines,ﬁeld emission displays and vacuum micro electronics.Field emitters have some important advantages over thermionic emitters:they have a higher brightness and lower energy spread,they can operate at ambient temperature and they have a lower power consumption because no heating of aﬁlament is required.Nevertheless,improvements are still desirable.For example,as it is shown above,the spatial resolution in low voltage electron probes is limited in part by the energy spread of theﬁeld emitter.If it would be possible to operate aﬁeld emitter at low voltage,battery driven applications are in reach(e.g. displays for laptop computers).The tunnel junction emitter is expected to combine the properties of low energy spread,high brightness,operation at low voltage and low power consumption.The tunnel junction emitter[2]is constructed by placing a sharp tip within tunneling range of a very thin metal foil(seeﬁg.9).Between tip and foil a voltage larger than the work function of the foil surface is applied.Provided that the foil is suﬃciently thin,a fraction of the tunneled electrons will travel through the foil without scattering.Electrons with suﬃcient forward energy to overcome the work function are emitted into the vacuum.In this way the work function acts as a high-pass energyﬁbined with the fact that the electrons originate from an atomic size tunneling area,a monochromatic high-brightness electron source is expected.As for most metals the work function is of the order of a few eV,the source is operated at low voltage.Although the emitted current is only a fraction of the tunnel current,the power consumption is still low because of the low voltage operation and because no heating is required.The emitter can be operated at high frequency because only a small voltage diﬀerence is needed to switch between on and oﬀand because the size of the emitter,and therefore its capacitance,can be kept small.This could be interesting for RF applications.As a tip is assumed to implement very sharp tungsten tip(often called nanotip)similar to that for STM investigations[11].However,the experiments with clean nanotip and free standing foil as it sketched inﬁg.9have shownthat thinﬁlm is damaging in a short time.That happens because of attractive forces between tip and foil.Fig.9To avoid this problem was proposed another[3]conﬁguration of experi-ments(seeﬁg.10).Fig.10.Cross section sketch of the device[3]Fig.11.TEM micrograph of9V dc etched oxide covered tungsten tip[11] As a tip was used an oxidized tungsten tip shown atﬁg.11.The experimental work has veriﬁed the principle of operating of this emit-ter.However,the stability and life time are not suﬃcient enough and still have to be improved.Nevertheless there is a hope that after some optimiza-tion of oxide layer,choice of proper tip material and development of reliable assembling technology would be possible to create a working device.It should be noted one more attractive property of solid state emitter that it is expected to be not that critical to vacuum condition as convenientﬁeld emitter.The low energy aberration correctorIn the basic form corrector is sketched inﬁg.12.Fig.12.Basic design of the foil corrector(not to scale).D:diameter of the aperture;s:gap between foil and aperture[4]It consists of aﬂat free-standing foil of nanometer size thickness with aper-tures on both sides.In the low-energy foil corrector,the foil is put on a retard-ing potential,such that the electrons have almost0eV kinetic energy when they enter the foil(and also when they have just left the foil at the other side).For use in a SEM additional optics is necessary to adjust correction and to focus the beam.A SEM column with corrector is shown atﬁg.13.Fig.13.SEM column with the foil correctorAs corrector is very strong negative lens it is necessary to put focussing lenses close to it.Because of that reason are favorable electrostatic lenses.Leaving apart all details of corrector calculations(see for details[12])we present aﬁnal result—a calculation example on realistic set-up,e.g.aberra-tion corrected low voltage SEM(seeﬁg.14).In the tab.2the measures of the design are given and electrode potentials and calculation result for optimum setting are given in the tab.3.Fig.14.Design of a low-voltage SEM column with a low-voltage foilcorrector[5].The design is rotationally symmetric in the z-axis Above the drawing,the numbering of the electrodes is designated.In the drawing,the paraxial ray for the settings in tab.3is shown.For the visibility, its radial extent is drawn5times larger than the maximum beam radius.Note that this ray has started inﬁn object position which is far left from the left border,thus it enters the column with a very small,but non-zero slope.Table2.Measures of the column design inﬁg.14.The design is mirror sym-metric in the foil,only the measures for the electrodes at the right side are listedelectrode no.aperture radius(mm)foil—rl0,30r21,00r31,00r41,00r51,00Table3.Optical properties calculated with aberration integrals for a foil po-tential of respectively0,1,0,4and1,0V.The potentials of the other electrodes with respect to the foil are the same as in tab.2optical ray tracing potential(V) property−394mm l5Z i12,3mm8900,1−0,275mm−1l3M−0,01552950,1−0,68mm l1C c1−0,83mm0,1rlC c25,3·102mm2950,1r39000,1r5of microcolumns,a lot of improvements has to be done yet.That concerns both construction of electron optical components and technology process for their embodiment.The detectorsFor transforming a microcolumn into SEM is necessary to equip it with appropriate detector of secondary and back scattered electrons(SE and BSE) as it sketched inﬁg.2.Besides small size the SE and BSE detector has to fulﬁll all common re-quirements:high collection eﬃciency,high gain at low voltages,fast response time and linearity in wide range of beam current.The most promising con-tender seems to be micro channel plate(MCP)and pin diode connected in a tandem manner.Particular construction of detector and technology process for its manufacturing and assembling with microcolumn has to be developed.The technology for microcolumn manufacturingSo far for microcolumn manufacturing were used or hybrid technology or MEMS technology.Each of those approaches has merits and demerits,which are well known to those skilled in the art,so it is not discussed here in details. We have being developing a technology similar to that in micro electronics, which seems to be more suitable for mass production.Have been developed technology process for lenses shown inﬁg.8.(seeﬁg.16and17).Also have been developed electrostatic octopole deﬂector–stigmator with thickness of electrodes about10µm,sketched inﬁg.18and19.Is assumed to implement two identical pieces placed between lenses in order to achieve deﬂection simultaneously with stigmation.ConclusionSince1990-th whenﬁrst miniaturized lenses have been micromachined, a substantial progress has been achieved in both methods for analytical com-putations of performance micro electron optics and prototyping of individ-ual electron optical elements.As for arrayed microcolumns,the only exam-ple of matrix4×4microcolumns for lithography purposes was presented by T.H.P.Chang and alii at2000-th.Nevertheless,in the foreseeable future one can expect the appearance of the systems manufactured by more advanced technology.By our opinion the progress in this area would be determined by tech-nology starting from manufacturing individual components and ending with assembling complete system.So when developing technology for any element as cathode,lens or detector is necessary to think from the beginning about its compatibility with whole technology process.Fig.16.Non symmetrical two electrodes lens designFig.17.100µm micro lens designFig.18.Design of octopole stigmator.Top view(a)and axial section(b)References[1]T.H.P.Chang,M.G.R.Thompson,M.L.Yu,E.Kratshmer,et all.Elec-tron beam microcolumn technology and application//SPIE.Vol.2522.P.4–12,(invited paper).[2]G.P.E.M.van Bakel,E.G.Borgonjen,C.W.Hagen,and P.Kruit.Cal-culation of the electron-optical characteristics of electron beams trans-mitted into vacuum from a sharp tip-thin foil junction//J.of Appl.Phys.Vol.83(1998)P.4279.[3]R.H.van Aken,M.A.P.M.Janssen,C.W.Hagen and P.Kruit.A simplefabrication method for tunnel junction emitters//Solid State Electron-ics.Vol.45.(2001)P.1033.[4]R.H.van Aken,C.W.Hagen,J.E.Barth and P.Kruit.Low-energy foilaberration corrector//Ultramicroscopy.Vol.93.(2002)P.321.[5]R.H.van Aken,C.W.Hagen,J.E.Barth and P.Kruit.Design of a low-voltage SEM equipped with the low-energy foil corrector//Submitted to Ultramicroscopy.[6]V.V.Aristov,V.V.Kazmiruk,V.A.Kudryashov,V.I.Levashov,S.I.Red’kin,C.W.Hagen,and P.Kruit.Microfabrication of ultrathin free-standing platinum foils//Surface Science.V.337.1998.P.402–404. [7]H.S.Kim,S.Ahn,D.W.Kim,Y.C.Kim and H.W.Kim,S.J.Ahn.Sub-60-nm Lithography Patterns by Low-Energy Microcolumn Lithog-raphy//Journal of the Korean Physical Society.Vol.49.P.S712–S715.[8]Ho-Seob Kim,Dae-Wook Kim,Seungjoon Ahn,Sung-Soon Park,Myeong-Heon Seol and Young Chul Kim,Sang-Kook Choi and Dae-Yong Kim.Multi-Beam Microcolumns Based on Arrayed SCM and WCM//J.of the Korean Physical Soc.Vol.45,No.5.P.1214–1217.[9]V.V.Kazmiruk,T.N.Savitskaja.arXiv:0805.0248v1[physics.ins-det].[10]J.E.Barth.Private communication.[11]V.V.Dremov,V.A.Makarenk,S.Y.Shapoval,O.V.Troﬁmov,V.G.Beshenkov and I.I.Khodos.Sharp and clean Tungsten Tips for STM investigation//Nanobiology.Vol.3.(1994)P.83–88.[12]R.H.van Aken,M.Lenc and J.E.Barth.Aberration integrals for thelow-voltage foil corrector//Nuclear Instruments and Methods in Physics Research A.Vol.519.(2004)P.205and Vol.527(2004)P.660.s。

六方氮化硼在二维晶体微电子器件中的应用与进展

第58卷第2期2021年2月徵鈉电子技术Micronanoelectronic TechnologyVol. 58 No. 2F eb ru ary2021D O I：10. 13250/j. cnki. wndz. 2021. 02. 003令材料与结构$六方氮化硼在二维晶体微电子器件中的应用与进展高渤翔U2，方茹3，吴天如1(1.中国科学院上海微系统与信息技术研究所信息功能材料国家重点实验室，上海200050;2.中国科学院大学，北京100049; 3•中国电子科技集团公司第三十二研究所，上海201808)摘要：六方氮化硼U-B N)因其优异的性能和潜在的应用前景而受到广泛关注。

着眼于/i-BN 在微电子器件领域中的发展与应用，总结了近年来国内外通过化学气相沉积（C V D)方法实现A-B N的高质量、大规模可控制备及图形化的代表性工作。

围绕A-B N的高介电常数、原子级平滑表面、高导热性和高稳定性，重点介绍了/i-B N在二维晶体介电衬底、半导体器件热管理平台以及集成电路封装材料中应用的研究进展，并简述了将A-B N应用于隧穿器件和存储阵列的研究成果。

最后，对A-B N在新型微电子器件大规模应用的已有成果进行总结，并展望了该领域未来的研究与发展方向。

关键词：二维（2D)材料；六方氮化硼（A-B N);微电子器件；介电衬底；隧穿器件；存储阵列；封装材料中图分类号：0612.3;TN305 文献标识码：A 文章编号：1671-4776(2021) 02-0107-07Application and Progress of Hexagonal Boron Nitride in Two-Dimensional Crystal Microelectronic DevicesGao Boxiang1’2, Fang Ru3，Wu T ia n ru1(1. State K ey Laboratory o f Functional Materials fo r Informatics ^Shanghai Institute o f Microsystem andInform ation Technology^ Chinese A cadem y o f Sciences^ Shanghai200050» C h in a；2. University o f Chinese A ca d e m y o f Sciences ^Beijing100049, China \3. The32n d Research Institute^China Electronics Technology Group Corporation ^Shanghai201808, China)A bstract：Due to excellent properties and potential application prospects, H exagonal boron nitride (/i-B N) has received extensive attention. Based on the developm ent and application of /i-BN in microelectronic devices, the recent representative studies of high-quality, large-scale controllable preparation and pattern in g of /i-BN achieved by chemical vapor deposition (C V D) m ethod at home and abroad are summarized. T he research progresses of /z-BN for tw o-dim ensional crystal dielectric s u b s tra te s, the therm al m anagem ent platform of sem iconductor devices and integrated circuit packaging m aterials are reviewed em phatically from the aspects of the high dielectric consta n t, atom ic level sm o o th surface, high therm al conductivity and high stability of /i-BN. T h e research results of the application of /i-BN in tunneling devices and m em ory arrays are briefly收稿日期：2020-08-17基金项目：中国科学院战略性先导科技专项（XDB3000()000)通信作者：吴天如，E-mail: *************微鈉电子技术presented. Finally, the obtained achievem ents of large-scale applications of /i-BN in new m icroelectronic devices are sum m arized, and the future research and trends of the field are prospected. Key w o rd s：two-dim ensional (2D) m a te ria l；hexagonal boron nitride (/i-B N)；microelectronic device；dielectric s u b s tr a te；tunneling device；memory a r r a y；packaging materialE E A C C：0500；2550()引百六方氮化硼U-BN)具有与石墨烯相似的二维蜂窝结构但完全不同的物理特性。

EVO 18 Product Specification

Carl Zeiss SMT - Nano Technology Systems DivisionP rod u ctSp e cifi c at ion产品技术规格EVO ® 18=Special Edition Scanning Electron Microscope扫描电子显微镜EVO® 18 - Product Specification产品技术规格产品技术规格For Materials Analysis, the EVO® 18 scanning electron microscope is a comprehensive high resolution imaging tool for a very wide range of applications. The EVO® 18 enables users toachieve previously unattainable image quality, and analytical accuracy, for non conductors using advanced variable pressure SEM technology. The large multi ported chamber includes one inclined port for EDS as standard. X-ray analytical techniques operate at the class leading analytical working distance of only 8.5 mm. Two horizontal ports are available forother accessories.对于材料分析，EVO® 18扫描电子显微镜是一台综合性的用途广泛的高分辨率成像工具。

对于非导体使用先进的可变压力SEM技术，EVO® 18允许用户获得以前难以得到的成像质量，并且分析准确。

Development of a Functional Sealing Layer for SOFC Applications

Development of a Functional Sealing Layerfor SOFC ApplicationsN.Caron,L.Bianchi,and S.Me´thout (Submitted June 26,2008;in revised form September 30,2008)Solid oxide fuel cells (SOFCs)are widely considered as an alternative solution to the decrease in fossilenergy consumption.However,to achieve high efﬁciency and long-term stability for a SOFC stack,it is essential to maintain a stable hermetic seal.To obtain efﬁcient air tightness between two SOFC cells,a solid seal composed of a ceramic matrix charged with glass particles has been developed.Atmospheric plasma spraying was selected to produce the solid seal because it can be used on a wide range of substrates of various natures and shapes.The developed seal was found to be solid,undistortable,and adhesive to its support at ambient temperature.The sealing properties were acquired when the SOFC was put into service:the glassy phase migrated into the peculiar plasma-sprayed microstructure of the ceramic matrix toward the interface leading to the air tightness of the deposit.The performance of the seal was:the leak rate observed at 7kPa was 0.43Pa L/s and,as a comparison,the requirement of the US Department of Energy is 0.5Pa L/s.Keywords ceramic matrix,glass compound,sealing,SOFC1.IntroductionSolid oxide fuel cells (SOFCs)are promising energy conversion devices that produce electricity by an electro-chemical reaction of a fuel gas,such as hydrogen or methane,with oxygen (Ref 1).Besides a highly efﬁcient process,advantages of SOFCs include low emission rates of air pollutants,the ability to use high-temperature exhaust for cogeneration or hybrid applications,and the ability of internal reforming (Ref 2).This family of fuel cells is mainly dedicated to high-power applications and mobile auxiliary power units.Whether in the form of a stack or individual cells,it is essential that the fuel and oxidant be kept separate from one another.Otherwise,the efﬁciency in producing ion exchanges across the electrolyte would be lowered and there would be a strong risk of exothermal explosions.Generally,the required sealant characteristics include thefollowing (Ref 3):coefﬁcient of thermal expansion well adapted to the other components to minimize thermal stresses and to improve thermal cycling behavior,sufﬁ-cient ﬂuidity to seal gaps at the sealing/assembly temper-ature,tailored viscosity at the operating temperature,gastight/marginal leak rate,thermal as well as chemical stability,electrical insulation,low cost,and ﬂexibility with regard to design.This generic set of requirements is not an exhaustive list because the characteristics of the sealant are also linked to the operating conditions of the stack.Both rigid and compressive seals are being developed to meet the challenge of the hermetic device,essential for the development of planar SOFC (Ref 4).A recent review by Fergus on this subject explained the differences between various seals and presented both advantages and drawbacks (Ref 5).A major advantage of compressive seals is the fact that they are not rigidly ﬁxed to the other SOFC components.Thus,an exact match of the thermal expansion is not required.However,a constant pressure must be applied to operate the gas-tight function during operation.For an effective compression,the seal must deform in response to the applied stress.So the main directions for the technical development consist in using ductile metals (Ref 6),materials of deformable shapes,or mica-based materials (Ref 7).On the other hand,rigid seals,which are mainly constituted of glass (Ref 8)and glass-ceramic sealants or metallic brazes,do not require this applied pressure.They have more stringent require-ments for adherence,cracking and thermal expansion matching.However,the main advantage of glass seals is the possibility to adapt the glass composition to get opti-mized physical properties (Ref 9,10).To go beyond the stress caused by differences in thermal expansion coefﬁ-cients,ﬁber-reinforced glass seals have been studied.Their advantages are a certain elasticity at the operatingThis article is an invited paper selected from presentations at the 2008International Thermal Spray Conference and has been expanded from the original presentation.It is simultaneously published in Thermal Spray Crossing Borders,Proceedings of the 2008International Thermal Spray Conference,Maastricht,The Netherlands,June 2-4,2008,Basil R.Marple,Margaret M.Hyland,Yuk-Chiu Lau,Chang-Jiu Li,Rogerio S.Lima,and Ghislain Montavon,Ed.,ASM International,Materials Park,OH,2008.N.Caron,L.Bianchi,and S.Me ´thout ,Atomic Energy Commis-sion,F-37260Monts,France.Contact e-mail:nadege.caron@cea.fr.JTTEE517:598–602Technical NoteDOI:10.1007/s11666-008-9248-31059-9630/$19.00ÓASM Internationaltemperature and a better compatibility with the thermal cycles of the SOFCs(Ref11,12).The main processes used for elaborating glass seals are screen printing or pneumatic spraying.In the case ofﬁber seals,a mechanical pre-compaction allows to elaborate auto-structured seals.In the literature,only a single report mentioning the use of plasma spraying could be found. Huang et al.(Ref13)investigated a multilayered com-posite seal structure consisting of thin layers of oxidation-resistant metals,porous ceramics,andﬁllers/glass as shown in Fig.1.The seal is an integrated part of the interconnected cells since it is directly coated onto the surfaces of matching parts owing to the use of atmospheric plasma spraying.During the stack assembly,the cells were joined through a heat/pressure-assisted curing process. The coating demonstrated an excellent mechanical robustness(thermal shock,tensile adhesion),high elec-trical resistivity,and low gas permeability.The seal developed at the Laboratory of Thermal Spraying in CEA(Monts)differed rather signiﬁcantly from those previously reported,and has led to a patent ﬁling(Ref14).The plasma spray technique(Ref15)was used to elaborate a coating composed of glassy particles dispersed into a matrix that remained solid during high-temperature operations.Details of elaboration as well as the mechanisms are presented in the next sections.Moreover,initial tests with regard to the leak rate and thermo-cycling resistance are reported.2.Experimental ProcedureThe materials were selected according to their coefﬁ-cients of thermal expansion(CTE),which is one of the fundamental characteristics for an efﬁcient seal.The ceramic matrix was ZrO2-8wt.%Y2O3(particle size range 5-25l m,Medicoat France).Whereas the glassy com-pound was a borosilicate(particle size range2-22l m, DMC2France S.A).Its detailed composition(supplier data)was B2O336wt.%,SiO231wt.%,Na2O16.7wt.%, CaO7.6wt.%,MgO3.6wt.%,and Al2O30.87wt.%.The properties of this glass are:half-sphere tempera-ture=710°C;softening point under pressure=580°C;and thermal expansion coefﬁcient=8.9910-6K-1.The half-sphere temperature is the temperature at which the sam-ple changes its shape so that its base radius equals its height,taking the shape of a half-sphere.Boron oxide is an important additive to silicate glasses since it decreases the glass viscosity,as well as the soft-ening point and glass transition temperature of the SOFC sealant(Ref16).The presence of calcium and magnesium in the C105is known to promote adherent and stable interfaces with yttria-stabilized zirconia(YSZ)(Ref17).The composite powder was manufactured by mechan-ically blending20wt.%glassy C105compound with ceramic YSZ powder.The choice of the plasma spraying process was dictated by the need for cost reduction com-pared to other manufacturing routes.Atmospheric plasma spraying(APS)was carried out with an automatic plasma system equipped with a Sulzer-Metco F4VB plasma gun (Wolhen,Switzerland),a turntable,and a six-axis robot. The plasma spray parameters were optimized with respect to the Ar/H2working gases,the spray distance,and the powder injection.The APS seal was characterized with scanning electron microscopy(SEM)in back-scattered electron(BSE)mode to differentiate the different components while the porosity and the glass percentage were measured by the combination of two techniques:Archimedes Porosimetry and Helium Picnometry.These volumetric methods involved the elaboration of free-standing samples.Adherends(e.g., stainless steel IC)PressurePressure filler infiltration& curingHermetic filler material(e.g., glass)t = 200~400micronsMetallic bond coat Stable corrosion resistantoxide (Al2O3) layer formednaturally during APS coatingPorous ceramic layer(top coat)Fig.1Scheme of the conceptual structure of the integrated composite seal(Ref13)3.Principles of the Plasma-Sprayed SealAs shown in Fig.2,the coating consisted of ﬂattened droplets with interlamellar contacts,microcracks,and globular porosities.The total void content of this typical APS coating is around 13%.The percentage of glass compound after plasma spraying is more important (up to 40%)than the one contained in the blending feedstock because of a heterogeneous treatment in the plasma jet.Unfortunately,the weak contrast difference does not allow highlighting this phenomenon.However,some glass inclusions could be seen forming dispersed within the matrix of YSZ droplets.In fact,during the SOFC operation,the glass com-pound softened whereas the matrix remained solid.Owing to a low viscosity and no reaction with the YSZ matrix,glass could migrate by capillarity through the three-dimensional network of interconnected cavities toward the free surface of the ceramic matrix involved in theglassy seal coating.At the same time,the glass ﬁlled most of the defects linked to the plasma spraying process.This gave rise to a continuous network and effectively blocked the continuous 3D leak path.Figure 3displays a SEM image of a cross section of plasma-sprayed seal after annealing during 1/2h at 900°C in air.Due to the poor optical contrast between the glass compound and the ceramic matrix,the plasma-sprayed microcracks that are usually ﬁlled with glass were no longer visible.The only thing that remained was a closed porosity linked to cavities that initially contained the glass compound,but were emptied due to the glass migration.The overall porosity remained constant.4.Characterization of the Plasma-Sprayed Seal4.1Leak Rate MeasurementsTo qualify leak rate of the plasma-sprayed seal,the ‘‘pressure loss’’method was chosen.This method allows the measurements at the SOFC operating temperature.The setup for the leak rate test was composed of a gas feeder,an insulation valve,a data recording system,a buffer volume,and an oven that was able to reach a temperature of 1200°C.The seal (thickness =150l m)was produced on a conical form corresponding to the tested design,on which a load of 19.6kPa was applied (Fig.4).The seal was heated to 900°C at 2.5°C/min.After the furnace reached the desired temperature,the inner part of the conical form was ﬂushed with nitrogen to perform the sealing tests for several pressure differentials as shown in Fig.5.Three domains were observed:[0-40kPa]:leak rate ranged between 0and 2.4Pa L/s;[40-70kPa]:leak rate was stabilized;[>50kPa]:leak rate showed a linear increase from 3.15to 5.94Pa L/s.According to the US Department of Energy,the requirement of a good sealing is 0.5Pa L/s under a differential pressure of 7kPa.The plasma-sprayed seals obtained in our work resulted in the leak rate of 0.43Pa L/s and are therefore suitable for application in SOFC.4.2Thermal CyclingThermal cycle testing was conducted by heating the specimen in air at a rate of 2.5°C/min to 900°C,keeping constant this temperature for 30min,and then cooling to ambient temperature,under a pressure differential of 20kPa,before reheating under the same conditions.No degradation of the leak rate was observed after four cycles.4.3Rupture Strength TestThe second application of this particular plasma-sprayed seal in the possibility is to join two interconnec-tors in a SOFC stack as shown in Fig.6.The seal was plasma sprayed onto the ﬁrst sand-blasted anddegreasedFig.2SEM image of a cross section of the plasma-sprayedsealFig.3Cross section of a plasma-sprayed seal after annealing at 900°C during 30min in airinterconnector (CROFER APU 22-Thyssen Krupp),and the second interconnector was placed on the free surface of the plasma-sprayed seal.At the conditioning tempera-ture (900°C)after a heating rate of 2.5°C/min,due to its viscous ﬂow,the glass compound ﬁlled the free space between the two pieces giving rise to an air-tight layer as shown in the top part in Fig.6.To quantify the rupture strength of the joining,a test was developed.It consisted of clamping the bonded assembly into a ﬁxture then increasing the air pressure until the interface,composed of the seal between the two interconnectors,breaks.The rupture pressure for a seal with a thickness of 220l m was measured to be around 250kPa.5.ConclusionsOne of the challenges for implementation of SOFCs relies in maintaining a hermetic sealing that remainsresistant and stable over the lifetime of the stack.To efﬁciently and hermetically join two SOFC repetitive units or components,a plasma-sprayed solid seal constituted of a ceramic matrix charged with glass particles was developed.This seal was found to be solid,undistortable,and bonded to its support at ambient temperature.It could also be realized as a self-supported body with a thickness from 50l m to several millimeters.6 mmPlasma-sprayed sealOvenBuffer volumeAcquisition system (temperature, pressure loss)55 mmGas feederInsulation valve50mbarFig.4Experimental setup for leak ratetestingFig.5Measured leak rate as a function of pressuredifferenceFig.6Optical micrography of a cross section of two surfaces (CROFER 22APU)joined by the migration of the glass com-pound from the plasma-sprayed sealEven though the initial structure of the plasma-sprayed seal was microcracked,the principle of the glassﬁller migration leads to an increase in the gas tightness of the matrix and to the possibility of sealing two surfaces owing to the glassy coating at the free interface.This seal was tested in a stack ofﬁve SOFC cells under normal operating conditions.The performance of the stack was nominal and no internal combustion between hydrogen and oxygen was observed. AcknowledgmentsThe authors wish to thank Mr.Bories and Mr.ToulcÕhoat for their contribution in the development of the seals and Dr G.Renouard-Vallet for her help in the testing of the plasma-sprayed seals.References1.N.Q.Minh,Ceramic Fuel Cells,J.Am.Ceram.Soc.,1993,76(3),p563-5882.A.Boudghene Stambouli and E.Traversa,Solid Oxide Fuel Cells:A Review of an Environmentally Clean and Efﬁcient Source of Energy,Renew.Sustain.Energy Rev.,2002,6,p433-4553.K.Scott Weil,C.A.Coyle,J.S.Hardy,and J.Y.Kim,AlternativePlanar SOFC Sealing Concepts,Fuel Cells Bull.,2004,2004(5), p11-164.A.Khandkar,S.Elangovan,and J.Hartvigsen,SOFC SystemDesign Interactions with Stack Fabrication Technology,Ceram.Trans.,1996,65,p263-2775.J.W.Fergus,Sealants for Solid Oxide Fuel Cells,J.PowerSources,2005,147,p46-576.V.A.C.Haanappel,V.Shemet,I.C.Vinke,M.Gross,Th.Koppitz,N.H.Menzler,M.Zahid,and W.J.Quadakkers,Evaluation of the Suitability of Various Glass Sealant-Alloy Combinations Under SOFC Stack Conditions,J.Mater.Sci.,2005,40,p1583-15927.M.Bram,S.Peckers,P.Drinovac,J.Monch,R.W.Steinbrech,H.P.Buckkremer,and D.Stover,Deformation Behavior andLeakage Tests of Alternative Sealing Materials for SOFC Stacks, J.Power Sources,2004,138(1-2),p111-1198.S.Simner and J.W.Stevenson,Compressive Mica Seals for SOFCApplications,J.Power Sources,2001,102(1-2),p310-3169.K.A.Nielsen,M.Solvang,S.B.L.Nielsen, A.R.Dinesen,D.Beeaff,and rsen,Glass Composite Seals for SOFCApplication,J.Eur.Ceram.Soc.,2007,27(2-3),p1817-1822 10.R.Wang,Z.Lu¨,C.Liu,R.Zhu,X.Huang,B.Wei,N.Ai,andW.Su,Characteristics of a SiO2-B2O3-Al2O3-BaO-PbO2-ZnO Glass-Ceramic Sealant for SOFCs,J.Alloys Compd.,2007, 432(1-2),p189-193posite Sealant Materials for Solid Oxide Fuel Cell,PatentUS6,271,158B2,7August200112.S.Taniguchi,M.Kadowaki,T.Yasuo,Y.Akiyama,Y.Miyake,and K.Nishio,Improvement of Thermal Cycle Characteristics ofa Planar-Type Solid Oxide Fuel Cell by Using Ceramic Fiber asSealing Material,J.Power Sources,2000,90,p163-16913.X.Huang,K.Ridgeway,S.Narasimhan,K.Reifsnider,andX.Ma,Application of Plasma Sprayed Coatings in a Novel Integrated Composite Seal for SOFCs,Building on100years of Success,B.Marple et al.,Ed.,May15-18,2006(Seattle,WA), ASM International,200614.Solid Seal Which is Obtained by Means of Thermal Spraying,Patent WO/2007/042505,19July200715.P.Fauchais,A.Vardelle,and B.Dussoubs,Quo vadis ThermalSpraying?,J Therm.Spray Technol.,2001,10(1),p44-6616.R.Zheng,S.R.Wang,H.W.Nie,and T.-L.Wen,SiO2-CaO-B2O3-Al2O3Ceramic Glaze as Sealant for Planar ITSOFC, J.Power Sources,2004,128(2),p165-17217.R.E.Loehman,H.P.Dumm,and H.Hofer,Evaluation of SealingGlasses for Solid Oxide Fuel Cell,Ceram.Eng.Sci.Proc.,2002, 23(3),p699-710。

计量经济学导论-ch16

© 2013 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.
© 2013 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.
© 2013 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.
Rank condition The first equation is identified if, and only if, the second equation contains at least one exogenous variable that is excluded from the first eq.
© 2013 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole or in part.

电子商务专业英语期中考试试题

xxxx学院期中测试课程：电子商务专业英语姓名：学号：班级：一、单词/词组英汉互译(本大题共30小题，每小题1分，共30分)。

transaction: 交易供应商：supplier wholesaler：批发商commission: 佣金原材料：raw material value chain：价值链specification: 规格虚拟的： virtual enrich：使丰富ultimate:最终的诈骗：fraud 排名：rankingsolution: 解决方案使命：mission considerable：相当大的stocking trading:股票交易盈利模式：revenue model 志愿者：volunteerunique: 唯一的购物车：shopping cart conversion rate：转化率boost: 提高执行： execute 开始生效：kick inbear with: 忍受目标市场：targeted market comprehensive: 综合的two-way interaction：双向互动努力、尽力： endeavor 宽带：broadband二、名词解释，请对以下名词/短句用英文进行解释说明（本题共6道，每个5分，共30分）。

1、electronic businessElectronic Business, commonly referred to as "e-commerce" or "e-business",may be defined as the utilization of information and communication technologies(ICT) in support of all the activities of business.2、B2BBusiness-to-Business (B2B)is a term commonly used to describe commerce transactions between businesses like the one between a manufactuer and a wholesaler or a wholesaler and a retailer,i.e.,and both the buyer and the seller are business entity.3、advantages of internet auctionConvenience flexibility economical to operate increased reach4、domain nameA unique address on the Internet, like .my, where clients or potential customers can find your business.5、affiliate marketingAffiliate marketing is one of the oldest forms of marketing, and Internet has brought new life to this old standby（备用品）. With affiliate marketing, influencers promote other people's products and get a commission every time a sale is made or a lead is introduced.6、SEMSearch engine marketing (SEM) is a form of Internet marketing that involves the promotion of websites by increasing their visibility in search engine results pages (SERPs) primarily through paid advertising.三、段落翻译，请把下面的英文段落翻译成中文（本题共1道，共10分）。

1、下载文档前请自行甄别文档内容的完整性，平台不提供额外的编辑、内容补充、找答案等附加服务。
2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
3、如文档侵犯您的权益，请联系客服反馈,我们会尽快为您处理(人工客服工作时间：9:00-18:30)。

Two Short Term Applications of the Semantic Web—Working Paper(Draft)Christoph WernhardPersist AGwernhard@March6,20011IntroductionWe outline two application scenarios which might be in the realm of short term applications of the Semantic Web:a software packaging system and the organization of a business trip.Both of them can be solved with today’s technology to some degree,so they do not show the novel potential of the Semantic Web in full.However considering Semantic Web solutions for them is useful to get a picture of the characteristics of the Semantic Web and become aware of some concrete technical issues.2A Software Packaging SystemWe take as a starting point the Debian GNU/Linux packaging system[1, 2],which successfully uses meta information on currently four thousands of independently maintained packages of bundled software to ensure their coordinated functioning.Roles and Sources of Meta Information In contrast to e.g.a keyword index,the meta information of a software packaging system must have a very high accuracy.Otherwise computer systems using it would be out of function very quickly.In the Debian system the meta information is provided in a distributed manner for each package by its maintainer,who should be the most compe-tent to determine it.This meta information is tested,in a sense,by the large user community that runs Debian systems in pre-release versions.•This has some similarities with the Semantic Web scenario in general:1–Information is structured and has high accuracy.1–Information is maintained independently by many sites,which are the most competent for doing that.–Such fragments of information are combined for larger tasks.–The required accuracy of information is validated continuously bya user community.Composition A software package is installed on a system or not,it is never consumed by other software that uses it.•A physical object,on the other hand,is consumed when it becomes part of a larger object,i.e.it is no longer available to become part of another object.•Therefore,for combination of software packages,the resource orienta-tion of inferences plays a less important role than for physical systems. Cost of Components Since Debian GNU/Linux consists of free software, the only cost of installing a software package are disk space,download time and maybe the human eﬀort needed for theﬁnal conﬁguration.•The packaging system could be extended for commercial software with price information.In contrast to the other attributes,a price is not an inherent property of a package,but also depends on the particular vendor.•Software vendors could independently provide pricing information about the packages,that can be combined with the other meta information. Identiﬁcation of Components A package is identiﬁed just by a symbolic name.•This is in contrast to more complex characterizations involving e.g.subtype relationships or shared instantiation of parameters.•However in some cases the package name itself contains meta informa-tion:A packages named xxx-doc contains documentation for package xxx,xxx-dev contains development libraries and headers for package xxx,task-xxx combines a set of other packages relevant to an applica-tion task(more on task-packages below).1Of course the Semantic Web can incorporate poorly structured information sources, but once information is going to be processed e.g.by an inference engine,highly structured extracts have to be used.2Description of the Semantics of Components The package meta in-formation attribute Description is a text description of the package.Along with the symbolic value Section(e.g.base,admin,graphics,editors,games) it provides information about the package’s semantics.Text search in the package descriptions can be used toﬁnd a package suited for a given appli-cation purpose.•One can imagine a knowledge based extension about the application relevant capabilities of a software package: e.g.ﬁle types it can han-dle,edit,display,convert,quality measures,certiﬁcations,background knowledge required by the user etc.•Third parties could independently provide such additional meta infor-mation.Component Dependencies Package dependencies are described with at-tributes that have an ordered and/or-structure of package names as argu-ment:Depends,Recommends and Suggests express decreasing levels of de-pendency.Depends means that the other package has to be installed for the ﬁrst to work properly.Conflicts means that both packages may not be installed at the same time.Replaces means that the other package has to be removed.The Debian installation procedure consists of two phases:unpacking and setting up a package.The latter step is called conﬁguration.Since the dependency constraints only regard conﬁguration,a set of packages can be downloaded and unpacked in an arbitrary ordering,e.g.one that ensures that CD-ROM disks have to be changed as less as possible.2After that the packages are going to be conﬁgured in their dependency ordering. Representation by Resource Oriented Inference Rules3•configured(p1),configured(p2),configured(p3):=action(configure(p1))unpacked(p1),configured(p2),configured(p3).•This rule expresses that package p1depends on p2and p3:the action configure(p1)leads the computer system from a state in which p1is unpacked and p2and p3are conﬁgured to a state in which p1is also 2A special attribute Pre-Depends can enforce dependency constraints at unpacking time.3See Section4for the syntax and intended meaning of the rules.3conﬁgured.configured(p2)and configured(p3)are repeated on the left side of the rule,since they are prerequisite for application of the action,but not consumed by it.The unpacked(p1)fact is consumed by the action,the configured(p1)fact is produced by it.•Some cases of Conflicts might be expressed by resource oriented facts:It is not possible to have two mail transport agents installed because both need to listen to the network to receive mail.This might be modeled by describing both as consuming the same network resource.•“Soft constraints”such as Recommends(no strong dependency but also chosen for installation by default),and Suggests(no strong depen-dency and not chosen for installation by default)might be important in practice.How can they integrated into the inference mechanism? Special Components A so called virtual package expresses an abstract capability.It has a symbolic name,that can appear as value of the Provides attribute,declaring that some package provides that abstract capability.A further package can then depend on the virtual package,which means that just one of the packages providing the virtual package has to be installed. An example for a virtual package is httpd(i.e.Web server),which is provided for example by packages apache,aolserver,boa,cern-httpd and roxen.Task-packages are virtual packages that combine by their dependents a set of packages for some larger application task chosen explicitly by the user, e.g.task-sgml,task-japanese,task-laptop,task-games.Processing The Debian installation program maintains information about the packages installed on the system as well as about the set of available packages by collecting package meta information from a set of user-speciﬁed sources such as CD-ROMs and FTP sites.If invoked with a set of package names the installation program determines which of the speciﬁed packages is not already installed,which dependents have to be installed and which conﬂicting packages have to be removed.It prints a description of what it would do,that has to be conﬁrmed by the user before it is going to be executed.•This has some similarity with the operation of a Semantic Web plan-ning agent.In one phase information is gathered from various sources, in a second it is combined into an execution plan that takes into ac-count a given“start”state.After user conﬁrmation the plan is going to be executed.44This is a conceptual base architecture only.In practice gathering,planning and exe-cution might happen in a more distributed and interleaved manner.4Alternative Compositions In the Debian packaging system,alterna-tives are simply resolved by preference according to the ordering of disjuncts in the attribute values.•The resource oriented inference approach allows the generation and comparison of diﬀerent possible plans.Atﬁrst sight this feature might not be needed for a software packaging system,but there are circum-stances under which it would be useful:•If software that costs something is included,diﬀerent installation plans might have essentially diﬀerent properties.•Disk space is not cheap on all systems.•Planning can contribute to“mental economy”:Combined with knowl-edge based package descriptions,one could determine software needed for certain tasks and get hints which new software systems must be learned to use.3Organization of a Business TripA business trip requires coordinated use of services of several diﬀerent kinds, oﬀered by several providers,where each kind of service again is provided by several competitors.This includes transport(walking,use of own car,taxi, rental car,local public transport,railway,airplane),hotels and restaurants. Composing and Comparing Services In today’s Web detailed infor-mation about most of these services as well as electronic booking facilities are available,such as timetables,route-planners,hotel and restaurant guides, price lists ually each of them has its very own graphical Web inter-face,that must be crawled through manually to get information needed to make decisions(pare prices,schedules,availability)and to imple-ment decisions(i.e.order services).•With the Semantic Web the coordinated composition of these services based on common interfaces is possible.These interfaces can be sup-ported by the service providers directly or by mediators.A planning agent computes diﬀerent compositions and compares them.After the user has committed to one of the suggested plans,the agent performs the necessary orderings and prints out the travel documents.•The Semantic Web further allows to directly connect service oﬀerings with information from the customer’s site.Business trip planning cane.g.–be directly connected to a personal date book;5–be used to organize a round trip with several meetings;–be used to organize a meeting involving several people,based on their date books.•A distributed transaction protocol,is required:if the failure of an order leads to the choice of a diﬀerent plan,previously made orders must be undone.The scope of transactions possibly should also include the planning phase.Existing Aggregations of Services There are some aggregations of the involved services in today’s Web:e.g.car rental brokers or car rental agencies giving discount related to the airlines of arrival.•These existing aggregations do not involve comparison of solution plans involving diﬀerent kinds of services(e.g.railway vs airplane).•They seem to rest on particular agreements between the involved par-ties.Pre-arranged aggregations are cheaper for the customer,sincee.g.car rental brokers might buy large contingents and pass this ad-vantage to the user.On the other hand they lead to cartel agreements and monopolies.•The Semantic Web seems to create a market,in which anyone can participate by publishing an oﬀer or inquiry.•However the Semantic Web also seems to require strong control mech-anisms,such as“trusted sites”.•The Semantic Web is able to perform the function of a car rental broker in a distributed manner.The“contingent”is implicit in the number of orderings.•By the possibility of including oﬀers of small enterprises,special cus-tomer needs can be satisﬁed,e.g.a small car rental agency in a remote place can be included.Enterprise Identity In the past years enterprises tried hard toﬁnd their identity within the world of connected computers by pushing huge advertis-ing budgets into the HTML-Web.•The identity of the involved enterprises is hidden to the user during the planning(i.e.decision making)process of the Semantic Web.6Flexibility Business trip processes seem not to change very much in the rmation could be improved,e.g.a route-planner that takes ex-pected traﬃc on the particular date into account,car navigation systems. Technical alternatives to business trips might become convenient,e.g.video-conferencing.•So maybe theﬂexibility oﬀered by the planning approach is not really useful in this example.The Semantic Web provides a large amount of ﬂexibility in the way plans,i.e.business processes,complex products or services,are composed:new ways of composition which are not pre-conceived as a whole can be inferred from the small partial plans provided by the individual oﬀerers.•A drawback of composing plans from small units is that there might easily be lots plans generated,which are useless in a ways that have not been“pre-conceived”,which means that it could be diﬃcult toﬁlter them out.•Besides inferencing plans,a library of plans can be used.It includes descriptions of processes and complex products involving several suc-cessive steps.It is organized in a distributed way,similar to schemas describing object sturcture(ontologies).When using those given plans, the Semantic Web agent acts more as a scheduler than as a planner. Representation by Resource Oriented Inference Rules5•%%trip_done combines that the business meeting is done%%and the person is at home.It can be used as%%"toplevel"goal.%%trip_done(person,place,t_from,t_to,home_place),business_meeting_done(person,place,t_from,t_to),at(person,home_place):=action(trip_finished)business_meeting_done(person,place,t_from,t_to),at(person,home_place).%%If the person is at the right time and place,the%%business meeting can be done.The person remains%%at that place.%%business_meeting_done(person,place,t_from,t_to),at(person,place,t_to)5See Section4for the syntax and intended meaning of the rules.7:=action(do_business_meeting(...))at(person,place,t_from).%%Transport by air.%%at(person,dest_airport,t_arrival):=action(do_flight(person,flight_no,date)) at(person,start_airport,t_departure).•The transport services consume and produce(besides cost and time) at(person,location)facts.The start state consists of an at/2fact (along with a point of time),the goal consists of an at/2fact express-ing that the traveler is back home again and a have done business fact.The have done business fact requires a third at/2fact with a certain time,without consuming it.•How time and cost are represented precisely?•Hotel and restaurant provide shelter and food facts,which need to be consumed if the trip goes overnight or over a certain amount of time.How this is represented?4Appendix:Inference RulesWe use the following syntax for a resource oriented inference rule:<produced_facts>:=action(<action>)<consumed_facts>.This is similar to a Prolog clause,with<produced facts>the head and<consumed facts>the body.In contrast to Prolog,the head may contain multiple literals,and within the arrow(Prolog’s:-),there is an action term action.Also in contrast to(the logical meaning of)Prolog, <produced facts>and<consumed facts>are not idempotent.The infor-mal meaning of such a rule is,that from a state in which all members of consumed facts hold,by applying action,a successor state can be reached in which consumed facts do no longer hold(the facts are removed from the state description),and consumed facts hold(are added to the state description).If a fact must be present for a rule to be applied,but is not consumed by rule application,it must appear in both<consumed facts> and<produced facts>.References[1]Ian Jackson et al.:Debian Packaging Manual[2]Jason Gunthorpe:APT User’s Guide8。