Minimum-entropy data partitioning using reversible jump Markov chain Monte Carlo
北京大学-物理化学-第2章-热力学第二定律
2.1 变化的方向性------不可逆性
除可逆过程外，一切变化都有一定的方 向和限度，都不会自动逆向进行。热力 学的不可逆过程。
各类变化过程的不可逆性不是孤立而是 彼此相关的，而且都可归结为借助外力 使系统复原时在环境留下一定量的功转 化为热的后果。
有可能在各种不同的热力学过程之间建 立起统一的普遍适用的判据，并由此去 判断复杂过程方向和限度。
热机效率(efficiency of the engine )
功功W与任,所另何吸一热的部机热分从之Q高c比传温值给(T称低h )为热温热源(T机吸c ) 热效热源率Qh.,,或将一称热部为机分热所转机作化转的为
换系数，用 表示。 恒小于1。
W Qh Qc
Qh
Qh
(Qc 0)
或
nR(Th
卡诺定理的意义：（1）引入了一个不等号 I R ， 原则上解决了化学反应的方向问题；（2）解决了热
机效率的极限值问题。
卡诺定理：
所有工作在同温热源与同温冷源之间的热 机，其效率不可能超过可逆机。 Carnot循环：第二定律发展中重要里程碑。
指明了可逆过程的特殊意义
原则上可以根据Clausius或Kelvin说法来判断一个过程的 方向，但实际上这样做是很不方便，也太抽象，还不能指 出过程的限度。Clausius从分析Carnot过程的热功转化关 系入手，最终发现了热力学第二定律中最基本的状态函 数——熵。
即ABCD曲线所围面积为 热机所作的功。
卡诺循环（Carnot cycle）
•根据绝热可逆过程方程式
： 过程2 T V 1 h2
T V 1 c3
过程4：
T V 1 h1
TcV4 1
热力学第二定律
第三章热力学第二定律前面，所学的热力学第一律，是以“能量守恒原理”为基础，建立了U和H两个热力学函数，通过对过程ΔU和ΔH的计算，解决了过程的热效应问题。
然而，在一定条件下，一过程能否自动进行，进行到什么程度，亦即，过程的方向和限度问题，第一定律无能为力，这恰恰是第二定律所要解决的问题。
人类经验表明：一切自然界的过程都是有方向性的。
大家都知道：自然界中存在朝一定方向自发进行的过程，例如：热自动从高温物体传向低温物体，直至两物体温度相等；气体自动地从高压区流向低压区，直至各处压力相同，相互接触的不同气体，总是自动的相互混合均匀；电流总是从高电流处流向低电流处直至各处电势相等：浓度不均匀的溶液，自动地变成浓度均匀一致。
等等，这些过程都是可以自动进行的，叫“自发过程”。
显然，一切自然界的过程都是有方向性及一定的进行限度。
从未发现哪一自发过程可自动恢复原状。
为什么自发过程的逆过程不能自动进行？这就是第二定律所要解决的中心问题—判断过程的方向和限度问题。
究竟什么因素决定自发过程的方向和限度？从表面上看，似乎不同的过程，有着不同的决定因素。
如，决定热传导方向和限度的是温度T；决定气体流动的是压力p；决定电流的是电势V；等等。
决定化学反应的是什么？这就要找出：决定一切自发过程方向和限度的共同因素，以此作为判断的共同根据。
寻找一切自发过程方向和限度的判据，这就要研究自发过程的共同特征，根据经验总结热功转化规律，找出反映自发过程本质特征的状态函数—S,以ΔS判断过程的方向和限度。
进而又S据判据在特殊条件下，推演出了A、G状态函数，从而，得到更方便更实用的判据ΔA、ΔG。
§3.1自发变化的共同特征—不可逆性前已述及，一切自发过程都是有方向性的，亦即，自发过程进行之后，系统不能自动恢复原状。
若要让其恢复原状，环境中有什么变化？若让环境也复原，需要什么条件？现举例说明。
1. 理想气体向真空膨胀过程。
这是一个自发过程，当气体向真空膨胀时，Q = 0，W = 0，ΔU=0，ΔT=0。
求解线性方程组稀疏解的稀疏贪婪随机Kaczmarz算法
大小 k̂ 。②输出 xj。③初始化 S = {1，…，n}，x0 = 0，
j = 0。④当 j ≤ M 时，置 j = j + 1。⑤选择行向量
ai，i ∈
{
1，…，n
}，每一行对应的概率为
‖a‖i
2 2
‖A‖
2 F
。
⑥
( | ) 确 定 估 计 的 支 持 集 S，S = supp xj-1 max { k̂，n-j+1 } 。
行从而达到加快算法收敛速度的目的。算法 3 给出
稀疏贪婪随机 Kaczmarz 算法。
算法 3 稀疏贪婪随机 Kaczmarz 算法。①输入
A∈ Rm×n，b ∈ Rm，最大迭代数 M 和估计的支持集的
大 小 k̂ 。 ② 输 出 xk。 ③ 初 始 化 S = {1，…，n}，x0 =
x
* 0
=
0。④
置
k
=
0
时，当
k
≤
M
-
1
时。⑤计算
( {| | } ϵk=
1 2
‖b
1 - Ax‖k 22
max
1≤ ik ≤ m
bik - aik xk 2
‖a
‖ ik
2 2
+
)1
‖A‖
2 F
（2）
⑥决定正整数指标集
{ | | } Uk =
ik|
bik - aik xk
2
≥
ϵ‖k b
-
Ax‖k
‖22 a
‖ ik
2 2
ï í
1
ï î
j
l∈S l ∈ Sc
其中，j 为迭代步数。当 j → ∞ 时，wj⊙ai → aiS，因此
基于稳定域边界的主导不稳定平衡点（BCU）法前提条件的验证
基于稳定域边界的主导不稳定平衡点（BCU）法前提条件的验证第24卷第6期2004年6月中国电机工程ProceedingsoftheCSEEV_01.24NO.6Jun.200402004Chin.Soc.forElec.Eng文章编号:0258-8013(2004)06-0035.05中图分类号:TM712文献标识码:A学科分类号:470.4051基于稳定域边界的主导不稳定平衡点(BCU)法前提条件的验证江宁强,宋文忠,戴先中(东南大学自动化研究所,江苏南京210096)ARESTlUCTIoNoNTIⅢPoWERSYSTEMBYTIⅢoRETICAL REQIJISITIoNSoFTIⅢBCUMETHoDJIANGNing—qiang,SONGW_en-zhong,DAIXian—zhong (AutomationInstituteofSoutheastUniversity,Nanjing210096,China) ABSTRACT:Anecessaryconditionfortheemploymentof BCUmethodinpowersystemtransientstabilityanalysisis derivedfromtheoreticalrequisitionsofthismethod.Basingon resultsofdifferentialtopology,itisprovedtllat.foranot- completestablepowersystem,conditionsofBCUmethod requiretheassociatedgeneralgradientsystembenot-completestable.Additionally,forageneralgradientsystem,ifallthe equilibriumpointsarehyperbolic,andthestablemanifoldsas wellastheunstablemanifoldsoftheunstableequilibriumpointsonthestabilityboundarysatisfythetransversality condition,thenthesystemmusthaveaSOUl"Cepoint.Thisgives asufficientconditionforthenot-completestabilityofthe generalgradientsystem,i.e.ithasnosourcepoint.This conditiononlymlatestothetypeofequilibriumpoints.These conditionsareverifiedontheWSCCfour-machinesystemand IEEE50-machinetestsystem.KEYWORDS:Powersystemstability:Transientstability; BCUmethod;CompletestabUity摘要t通过对基于稳定域边界的主导不稳定平衡点法(Boundaryofstabilitybasedcontrollingunstableequilibrium pointmethod,BCU)的前提条件的分析,得到了当故障清除后的电力系统不完全稳定时,应用该方法的一个必要条件: 相关的广义梯度系统不完全稳定.并证明了使该条件得到满足的一个充分条件:广义梯度系统无源点.在稳定性分析中,可以通过检验该条件来间接地检验电力系统是否满足BCU法的前提条件.对一个4机系统和IEEE50机测试系统的计算验证了上述的结果.关键词:电力系统稳定性;暂态稳定性;BCU法;完全稳定性基金项目:国家杰出青年科学基金项目(59925718).1引言基于稳定域边界的主导不稳定平衡点法(Boundaryofstabilitybasedcontrollingunstable equilibriumpointmethod,BCU)是电力系统暂态稳定分析中的重要方法,它借助势能界面(PEBS)来确定主导不稳定平衡点(CUEP),并用CUEP处的暂态能量来确定故障后系统的临界能量.BCU法以严格的理论分析为基础【卜4】,因而受到普遍关注.但该法的部分理论前提无法直接检验.其中的单参数横截性条件是重要的理论前提,对单参数族而言,这不是一个具有通用性的条件【.文【6】通过实例分析说明:当该条件不满足时,BCU法有可能得不到正确的CUEP.因此,怎样检验实际的电力系统是否满足这个抽象的数学条件,是提高该法的可靠性所需解决的问题.本文通过对BCU法的前提条件的分析,得到了当故障清除后的电力系统不完全稳定时,应用该方法的一个必要条件.并证明了使该条件得到满足的一个充分条件.在稳定性分析中,可以通过检验该条件来间接地检验电力系统是否满足BCU法的前提条件.2BCU法的理论前提对电力系统的限制2.1BCU法的理论前提BCU法理论【】引入了单参数族,,旦一a,,一J一¨厂●●--J),纠;{}a.一=.36中国电机工程第24卷式中参数∈[0,1],该式描述了参数厮对应的一族系统,为势能函数【】;别为,z一1维和n维向量.该理论证明了如下定理:定理1[31如果对∈[0,1],单参数族中的系统应满足如下条件:(1)平衡点是双曲的;(2)稳定域边界上平衡点的稳定流形与不稳定流形满足横截性条件;(3)当t--->oo时,稳定域边界上每条轨线趋于一个平衡点.那么,系统0)和1)的稳定域边界上的平衡点相同.1)在n一1维角度空间中的子系统是一个广义梯度系统,记为G.定理l中的3个条件是BCU 法理论的前提条件,其中条件(2)又称为单参数横截性条件.下面运用微分动力系统原理来分析这些条件对电力系统的限制.2.2单参数族中系统的结构稳定性2.2.1紧流形上的单参数族微分动力系统理论【'研究了动力系统的结构稳定性,得到了系统结构稳定的充分条件,即公理A和横截性条件.公理A【.紧致的n维流形上的向量场满足:(1)非游荡集是双曲的;(2)周期点在非游荡集中稠密.定理2【紧致的n维流形上向量场若满足:(1)公理A;(2)所有稳定流形和不稳定流形的交点满足横截性条件.则该向量场是结构稳定的.需注意的是,上述条件仅限于紧流形上的系统.单参数族中的系统都定义在R2,.上,而R2,.不是紧流形,因此不能直接用上述的数学理论来讨论这些系统的结构稳定性.下面将引入一个覆迭映射,它将单参数族中的系统映射到紧流形上.在该紧流形上,可以进一步讨论系统的结构稳定性.文[10]指出,对有阻尼的电力系统,其频率偏差最终是有界的.对单参数族中的系统,记max.{-喜=mindi}由单参数族的定义可见,对系统中从('56,)出发的轨线[,以],若(>‰,则(Oi(t)<0;若(<一,则啦(f)>0.因此,中满足II≤M的区域是一个不变集,且任一轨线都将进入该区域.定义:区间=[-wwf]'wFk/,1.2….,n积流形=WIxw:x…xw.可得流形一xW".它包含了系统的非游荡集,可以在其中研究系统的性质.R?xW"中系统的向量场在各方向上周期性重复,即对点p=(,….,_l',….,)ER".×和点p+v处的向量是相同的,称这些点p+v为重复点.这里,v=27c×(v.,v2,…,v,0,0….,0),vl,v2….,v是任意整数.将流形,_[0,2x]中的点0和点27c定义为等同点[11】, 即可定义n个流形,的积流形,一.定义覆迭映射g为RxW一,¨xW(1)g-lI(,,…,一l,l,2,…,)=(l,2,…,一1,l,2,…,)(2)=mod2x,k=-I…2…n—l(3)将RxW"上的系统映射到紧流形P.×上,且P.xW"上的向量场至少是C的【….因此,单参数族中的系统都可由覆迭映射映射到紧流形上.下面借助于这个紧流形来讨论单参数族中系统的性质.2.2.2完全稳定电力系统及单参数族中系统的性质将每条轨线都收敛至平衡点的电力系统被称为完全稳定的电力系统【】,这个定义可推广至流形上的动力系统.定义1对流形上的动力系统,如果每条轨线都收敛至平衡点,则称该系统是完全稳定的.反之,如果至少有一条轨线不收敛至平衡点,则称该系统是不完全稳定的.对单参数族(中完全稳定的系统,可以证明下面的2个性质.性质1任一不稳定平衡点,必在某个稳定平衡点的稳定域边界上.证明:若系统完全稳定,则U:RxRn,xsEERxR为闭集;而UA()为开集.这里,I∈Et是稳定平衡点的集合,故RxR一UA()=jEEsUaA().对任一不稳定平衡点,由第6期江宁强等:基于稳定域边界的主导不稳定平衡点(BCU)法前提条件的验证37X..仨UA(x,),必有X..∈UOA(x,).j∈EsxjEE性质2对于完全稳定的系统,若在某种扰动下始终满足定理1中的条件(1)～(3),则系统在该扰动下保持完全稳定.证明:.xR"中的系统轨线都将进入.×,只需证明系统在R×上具有该性质.g将J×上的系统映射到紧流形P.×上.P×中过点P的轨线对应于.×中过P的重复点的一族轨线..×中,作为重复点的平衡点及其稳定和不稳定流形,在PJ×中共有同一个映像.因此,若系统在J×中完全稳定,则在,'×中的映像也完全稳定,反之亦然.此外,对.×上完全稳定的系统,g将J×中的双曲平衡点映射为PJ×中的双曲平衡点.对于双曲平衡点的稳定流形和不稳定流形,xW'中有横截交点的稳定流形A和不稳定流形在P.×中的映像仍横截相交;J×中无交点的稳定流形A和不稳定流形,如不含重复点,它们在PJ×中的映像必无交点.因此,若系统满足定理1中的条件(1)～(3),则映像系统也满足定理1中的条件(1)～(3).对于紧流形P.×中完全稳定的系统,若满足定理1中的条件(1)～(3),由定理2知,它是结构稳定的.因此,如果在某种扰动下,R×中的系统始终满足定理1中的条件(1)～(3),则系统在该扰动下保持完全稳定.2.3应用BCO法的必要条件电力系统的严重故障如果不及时清除,系统往往会失去同步.由BCU法的理论前提,可得使用BCU法的一个必要条件为:系统0)不完全稳定时,广义梯度系统G不完全稳定.证明:参数的连续变化可视为对单参数族中系统的扰动,由单参数族中完全稳定系统的性质1和性质2可知:系统0)完全稳定当且仅当系统1)完全稳定.1)的完全稳定性与广义梯度系统G的完全稳定性等价,因此该条件成立.可见,当电力系统0)不完全稳定时,通过检验广义梯度系统G是否不完全稳定,可以判断是否可以用BCU法来分析系统的暂态稳定性.如果此时的广义梯度系统G完全稳定,则电力系统0)不满足该法的理论前提.这样就将横截性条件的检验转化为广义梯度系统G的不完全稳定性的检验问题.3广义梯度系统不完全稳定性的检验与BCU法理论中的横截性条件相比,广义梯度系统的不完全稳定性较为直观.如果通过仿真得到系统G的一条不稳定轨线,就可以确定广义梯度系统是不完全稳定的.反之,由于不可能遍历整个系统,因此,如果用仿真方法得到了有限条稳定轨线,并不能确定梯度系统是完全稳定的. 下面引入稳定域集合的局部有限性的定义,并讨论广义梯度系统G的性质.定义2【l】记{A()}为中所有稳定平衡点稳定域的集合,其中A(.)表示一个稳定平衡点的稳定域.如果该集合中任意的点存在一个邻域,它与有限个稳定域相交,则称该集合为局部有限的.定理3若系统G完全稳定,则{A()}局部有限.文[13】中定理5.8,5.9的证明同样适用于该定理的证明,证略.定理4{A()}局部有限的充要条件是各稳定平衡点的稳定域有界.证明:①必要性.用反证法证明.若稳定平衡点的稳定域A()无界,则在A()中存在序列{},满足lin一8m=oo,且{mod2/r}收敛至R中的一点.由广义梯度系统的相图的周期性,可知{}中的点分别属于R?中无限多个不同的稳定域,故点的任一邻域必与无限多个稳定域相交,即{A()}非局部有限.因此,如果{A()}局部有限,则各稳定平衡点的稳定域有界.②充分性.等价地需要证明,如果{A()}局部无限,那么至少有一个稳定平衡点的稳定域是无界的.将区间,_[0,27c】中的点0与2兀定义为等同点,可得,z一1个流形, 的积流形,.这是一个紧流形【5】.广义梯度系统在,.上平衡点数的上限可用Morse—Smale不等式描述【钔.因此,稳定平衡点数r是有限的.对R中任意的一点x=-(4,'5l….,..),将点X+V定义为点X的重复点,v=2rcx(vl,v2….,.1),vl,v2,…,lPn.1,是任意整数.将所有的重复点定义为一个等价类,则R中所有稳定平衡点分属于,上,.个等价类(),4为稳定平衡点,i=1,2,...',.同样可定义稳定域的等价类(A().若{A()}不是局部有限的,则存在点yEA(国∈{A()},它的任意一个邻域与无限个稳定域相交.任取一个有界邻域U(y),在V(y)nA(国中各取一个点构成一个无穷序列{Yf},YiEA(,i=1,2….,oo.若yl∈A(),yj~A(,38中国电机工程第24卷,i=1,2….,oo,且,属于同一个等价类,则将Y,Y1列入一个子序列,这样,可由{Y)构造出k个子序列,j=l,2….,k,足≤,..这些子序列构成无穷序列{Y)的一个划分,因此,至少有一个子序列是无穷子序列.记一个无穷子序列为{., ….,),1≤p≤足,Ypf∈A(f),1,2….,oo.证明:在由稳定平衡点,i=1,2….,oo构成的序列{) 中,对于平衡点.=【lIl,.I2'…,..】,必有maxI一8plI=oo.若不然,设maxIf一ll=L<oo,则必有不大于7c的非负整数m,户1,2,...,,z一1,使得1f,f一l,『I27c×,,≥2,1≤.7≤,z一1,由此可知,序列中的平衡点个数小于等于H(2m,+1),与{)为无穷序列相矛盾.因此,l_<j<_n-i序列{)中,对于.,至少在一个角度方向上,设为方向b,1≤易≤,z一1,有m…axI一l,6l_oo.由系统向量场的周期性可知,)I(l一)∈A(6p1),因此A(6p1)无界.其充分性得证.定理5如果系统G满足:(1)平衡点是双曲的;(2)稳定域边界上平衡点的稳定流形和不稳定流形满足横截性条件;(3)存在一个稳定平衡点,其稳定域有界.则系统必有源点.证明:文【14】的定理4证明,如果系统dx/dt=-f(x)存在满足如下条件的能量函数:(1)非平衡点处ddf≤O;(2)沿从非平衡点出发的轨线,,集合{tER,dv(),df=O)在中的测度为O;(3),)有界则,有界.那么,如果一个稳定平衡点的稳定域有界,则该的能量函数,因此,如果存在一个稳定平衡点,它的稳定域有界,那么系统必有源点.由定理3—5,可得广义梯度系统G不完全稳定的一个充分条件为:广义梯度系统无源点.这个结果可用于判别广义梯度系统的不完全稳定性.4示例BCU法的理论前提是获得正确CUEP的充分条件.在现有理论的基础上,确定该法的适用范围有助于提高分析结果的可靠性.本文提出的必要条件反映了该理论的前提对电力系统的限制.例1表1是一个4机系统的计算结果【l5】.系统为均匀阻尼(DctM~O.15,i=1,2,3,4),发电机节点1为平衡节点.表中列出了在不同的发电机注入功率下,BCU法得到的CUEP.其中,故障投影轨线在梯度系统中的溢出点至不稳定平衡点的轨线用阴影法【16校正.与各个故障相对应的正确的CUEP采用文[17】的方法计算,并已通过仿真方法加以验证.在例1中,当梯度系统不完全稳定时,BCU法得到了正确的CUEP;当梯度系统完全稳定时,其结果可能不正确.此外,当梯度系统不含源点时,该梯度系统不完全稳定.例2BCU法对IEEE50机测试系统【l8】的分析结果与仿真结果基本吻合[3,19,20】.经验证,对于三相金属性对地短路故障,各节点的N一1故障对应的梯度系统都是不完全稳定的.如节点7处发生三相对地短路故障,切除支路6—7,清除故障后,电力系统不完全稳定.同时,该系统对应的梯度系统也不完全稳定.因此,该系统满足1.3节中的必要条件.BCU法得到的CUEP如表2所示,其中f-能量函数必在一个源点上达到最大值.1,2,3,4,为发电机内节点号,n=50,角度蹦单位对广义梯度系统G,势能函数是满足要求为o).表中结果与基准算法的结果是一致的.表1四机系统示例Tab.1Somecasesofafour-machinesystem第6期江宁强等:基于稳定域边界的主导不稳定平衡点(BCU)法前提条件的验证395结论通过对BCU法理论的分析可知,如果故障清除后的电力系统不完全稳定,用BCU法分析系统的暂态稳定性时,要求相关的广义梯度系统不完全稳定.对一个4机系统和IEEE50机测试系统的计算验证了上述的结果.本文还证明,没有源点是广义梯度系统不完全稳定的充分条件.这个条件只与平衡点的类型有关.参考文献【1】ChiangHD,WuFF,V araiyaPPABCUmethodfordirectanalysisof powersystemtransientstability[J】.IEEETrans.PowerSyst.,199l4,8(3):1194.1208.【2】ChiangHD.Analya~resultsondirectmethodsforpowersystem transientstabilityanalysis[C].AdvancesincontrolandDynamical Systems,V olumeXL,NewY ork:Academic,1991,43(3):275—334.【3】ChiangHD,ChuC.TheoreticalfoundationoftheBCUmethodfor directstabilityanalysisofnetwork-reductionpowersystemmodels withsmalltransferconductanee[J】.mEETram.onCircuitsand systems,1995.42(5):252—265[4】TongJ,ChiangHD,ConneenTPAsensitivity-basedBCUmethodfor fastderivationofstabilitylimitsinelectricpowersystems[J].IEEETrans.onPowerSystems,1993.8(4):1418—1428.【5】A.Llamas,J.DeLaReeLopez,i,eta1.Clarificationsofthe BCUmethodfortransientstabilityanalysisⅢ.IEEETransonPower Systems,1995,10(1):210-219.【6】PaganiniELesieutreBC.Geneticproperties,one—pm'mneter deformations,andtheBCUmethod【J1.IEEETramonCircuitsand Systems一1:1999,46(6):76763.[7】ChillingworthDRJ.Differentialtopologymaviewtoapplications[M].London:PitmanPublishing,1976.【8】Smale,S.Differentinbledynamicalsystems[J].Bul1.A.M.S.,1967,73: 747—817.【9】Robinson,R.C.StructuralstabilityofCflows[C].inDynamical Systems—Warwick1974~notesinMathematics468,ad.Manning,A.K.,Springer-V erlag,Berlin,1975.【10】AropostathisA,SastrySS,V araiyaPGlobalanalysisofswing dynamies[J].Ⅱ强lETramonCircuitsandSystems,1982,CAS一29(10):673—679.【11】【12】【13】【14】【15】【16】【17】叶彦谦.曲面动力系统【M】.科学出版社,1991. ArapostathisA.V araiyaPThebehaviorofthreenodepower networks[J].Int.J.Elee.PowerEnergysyst.,1983,5(1):22—30. ZaborszkyJ,HuangGZhengB,eta1.Onthephaseportraitofaclass oflargenonlineardynamicalsystemssuchasthepowersystem[J]. IEEETram.onAutomaticControl,1988,33(1):4-15ChiangHD,WuFF.V amiyaPFoundationsofdirectmethodsfor powersystemtransientstabilityanalysis[J].IEEETramonCircuits andSystems,1987,CAS一34(2).-160-172.傅书遏,薛禹胜,倪以信.直接法稳定分析【M】.中国电力出版社.1999.TreinenRT.VittalV,Kliemann,】v.Animprovedtechniqueto determinethecontrollingunstableequilibriumpointinapower system[J].IEEETramonCircuitsandSystems,1996,43(4):313—323. 王成山,贾宏杰,江小东(WangChengshan,JiaHongjie,Chiang Hsiao-Dong).一种寻找正确相关不稳定平衡点的基准算法(Standardmethodforfindingcorrectcontrollingunstableequilibrium point)[J].天津大学(JournalofTianjinUniyersity),1999,32(5):529—534.【18】IEEECommitteePeport.Transientstabilitytestsystemsfordirect stabilityataalysis[J].IEEETramonPowerSystems,1992,7(1);37-43.【19】曾沅,余贻鑫(ZenYuan,YuYLxin),电力系统动态安全域的实用解法(Apracticaldirectmethodfordeterminingdynamicsecurityregionsof electricpowersystems)[J].中国电机工程(ProceedingsoftheCSEE),2003,5:24—28.【20】杜正春,甘德强,刘玉田,等(DuZhengchun,GanDeqiang.Liu Yutian,et口D.电力系统在线动态安全评价的一种快速数值积分方法fAfastnumericalintegrationmethodforpowersystemon—line dynamicsecurityassessment)[J].中国电机工程(Proceedingsof theCSEE),1996,1:29—32.收稿日期:2004-01—08.作者简介:江宁强(1970-),男,博士研究生,研究方向为电力系统的稳定性及控制;宋文忠(1936-),男,教授,博士生导师,主要研究方向为系统辨识,复杂系统的控制和DEDS,戴先中(1954--),男教授,博士生导师,主要研究方向为神经网络,非线性控制,计算机控制,机器人控制,电力系统控制.(责任编辑喻银凤)。
3D电极的介电泳力与惯性力的粒子连续分选仿真
２０２２年第４１卷第３期 传感器与微系统（ＴｒａｎｓｄｕｃｅｒａｎｄＭｉｃｒｏｓｙｓｔｅｍＴｅｃｈｎｏｌｏｇｉｅｓ）ＤＯＩ：１０．１３８７３／Ｊ．１０００—９７８７（２０２２）０３—００４３—０４３Ｄ电极的介电泳力与惯性力的粒子连续分选仿真李晓红１，２，张斌珍１，段俊萍１，王佳云１，屈 增１，冀苗苗１（１．中北大学仪器科学与动态测试教育部重点实验室，山西太原０３００５１；２．太原工业学院电子工程系，山西太原０３０００８）摘 要：细胞分选在生物医学中起着重要的作用，而其中介电泳分选由于其无需生物标记，对粒子损伤小等优势得到了广泛的应用。
本文设计了一种结合三维（３Ｄ）电极的介电泳力和收缩—扩张结构的惯性力的微流控芯片，通过ＣＯＭＳＯＬＭｕｌｔｉｐｈｙｓｉｃｓ仿真软件对流体的流速分布、电场分布及粒子的运动轨迹进行仿真分析。
仿真结果表明：三维电极相较于传统的平面电极，能够提供粒子垂直运动方向上的非均匀电场，更有助于实现粒子的高效率分选。
此外，当粒子随流体运动时，不同的流道收缩—膨胀比分选效果不同，当收缩—膨胀比为３6１时，分选效果会更好。
通过仿真证明了所设计结构的有效性，确定了芯片尺寸，为后续的粒子连续高通量高效率分选，提供了重要的参考价值。
关键词：微流控芯片；惯性力分选；介电泳力分选；三维电极中图分类号：ＴＰ２１２．３ 文献标识码：Ａ 文章编号：１０００—９７８７（２０２２）０３—００４３—０４Ｐａｒｔｉｃｌｅｃｏｎｔｉｎｕｏｕｓｓｅｐａｒａｔｉｏｎｓｉｍｕｌａｔｉｏｎｂａｓｅｄｏｎ３ＤｅｌｅｃｔｒｏｄｅｃｏｕｐｌｅｄｗｉｔｈｄｉｅｌｅｃｔｒｉｃｅｌｅｃｔｒｏｐｈｏｒｅｓｉｓｆｏｒｃｅａｎｄｉｎｅｒｔｉａｆｏｒｃｅＬＩＸｉａｏｈｏｎｇ１，２，ＺＨＡＮＧＢｉｎｚｈｅｎ１，ＤＵＡＮＪｕｎｐｉｎｇ１，ＷＡＮＧＪｉａｙｕｎ１，ＱＵＺｅｎｇ１，ＪＩＭｉａｏｍｉａｏ１（１．ＫｅｙＬａｂｏｒａｔｏｒｙｏｆＩｎｓｔｒｕｍｅｎｔａｔｉｏｎＳｃｉｅｎｃｅａｎｄＤｙｎａｍｉｃＭｅａｓｕｒｅｍｅｎｔ，ＭｉｎｉｓｔｒｙｏｆＥｄｕｃａｔｉｏｎ，ＮｏｒｔｈＵｎｉｖｅｒｓｉｔｙｏｆＣｈｉｎａ，Ｔａｉｙｕａｎ０３００５１，Ｃｈｉｎａ；２．ＤｅｐａｒｔｍｅｎｔｏｆＥｌｅｃｔｒｏｎｉｃＥｎｇｉｎｅｅｒｉｎｇ，ＴａｉｙｕａｎＩｎｓｔｉｔｕｔｅｏｆＴｅｃｈｎｏｌｏｇｙ，Ｔａｉｙｕａｎ０３０００８，Ｃｈｉｎａ）Ａｂｓｔｒａｃｔ：Ｃｅｌｌｓｅｐａｒａｔｉｏｎｐｌａｙｓａｎｉｍｐｏｒｔａｎｔｒｏｌｅｉｎｂｉｏｍｅｄｉｃａｌａｐｐｌｉｃａｔｉｏｎｓ．Ｄｉｅｌｅｃｔｒｏｐｈｏｒｅｓｉｓｉｓｗｉｄｅｌｙｕｓｅｄｄｕｅｔｏｉｔｓａｄｖａｎｔａｇｅｓｏｆｎｏｒｅｑｕｉｒｅｍｅｎｔｆｏｒｂｉｏｌｏｇｉｃａｌｍａｒｋｅｒｓａｎｄｎｏｄａｍａｇｅｃａｕｓｅｄｔｏｐａｒｔｉｃｌｅｓ．Ａｍｉｃｒｏｆｌｕｉｄｉｃｃｈｉｐｃｏｍｂｉｎｅｄｗｉｔｈｔｈｅｅｌｅｃｔｒｏｐｈｏｒｅｔｉｃｆｏｒｃｅｏｆｔｈｅ３Ｄｅｌｅｃｔｒｏｄｅａｎｄｔｈｅｉｎｅｒｔｉａｆｏｒｃｅｏｆｔｈｅｃｏｎｔｒａｃｔ ｅｘｐａｎｓｉｏｎｓｔｒｕｃｔｕｒｅｉｓｄｅｓｉｇｎｅｄ．ＣＯＭＳＯＬＭｕｔｉｐｈｙｓｉｃｓｓｉｍｕｌａｔｉｏｎｓｏｆｔｗａｒｅｉｓｕｓｅｄｔｏａｎａｌｙｚｅｔｈｅｆｌｏｗｆｉｅｌｄ，ｔｈｅｅｌｅｃｔｒｏｄｅｆｉｅｌｄａｎｄｐａｒｔｉｃｌｅｔｒａｊｅｃｔｏｒｙ．Ｔｈｅｓｉｍｕｌａｔｉｏｎｒｅｓｕｌｔｓｓｈｏｗｔｈａｔｃｏｍｐａｒｅｄｗｉｔｈｔｒａｄｉｔｉｏｎａｌｐｌａｎａｒｅｌｅｃｔｒｏｄｅｓ，ｔｈｅ３Ｄｅｌｅｃｔｒｏｄｅｓｃａｎｐｒｏｖｉｄｅｎｏｎ ｕｎｉｆｏｒｍｅｌｅｃｔｒｉｃｆｉｅｌｄｉｎｖｅｒｔｉｃａｌｄｉｒｅｃｔｉｏｎ，ｗｈｉｃｈｉｓｍｏｒｅｕｓｅｆｕｌｔｏｒｅａｌｉｚｅｅｆｆｉｃｉｅｎｔｐａｒｔｉｃｌｅｓｅｐａｒａｔｉｏｎ．Ｉｎａｄｄｉｔｉｏｎ，ｗｈｅｎｔｈｅｐａｒｔｉｃｌｅｓｍｏｖｅｗｉｔｈｔｈｅｆｌｕｉｄ，ｔｈｅｓｅｐａｒａｔｉｏｎｒｅｓｕｌｔｓａｒｅｄｉｆｆｅｒｅｎｔｗｉｔｈｄｉｆｆｅｒｅｎｔｃｏｎｔｒａｃｔｉｏｎ ｅｘｐａｎｓｉｏｎｒａｔｉｏ．Ｔｈｅｓｏｒｔｉｎｇｅｆｆｅｃｔｉｓｂｅｔｔｅｒｗｉｔｈｗｈｅｎｔｈｅｃｏｎｔｒａｃｔｉｏｎ ｅｘｐａｎｓｉｏｎｒａｔｉｏｉｓ３6１．Ｔｈｅｄｅｓｉｇｎｅｄｓｔｒｕｃｔｕｒｅｉｓｐｒｏｖｅｄｅｆｆｅｃｔｉｖｅｔｈｒｏｕｇｈｔｈｅｓｉｍｕｌａｔｉｏｎｓ，ａｎｄｔｈｅｓｉｚｅｏｆｔｈｅｍｉｃｒｏ ｄｅｖｉｃｅｉｓｃｏｎｆｉｒｍｅｄ．Ｔｈｉｓｓｔｒｕｃｔｕｒｅｐｒｏｖｉｄｅｓａｎｉｍｐｏｒｔａｎｔｒｅｆｅｒｅｎｃｅｖａｌｕｅｆｏｒｃｏｎｔｉｎｕｏｕｓｈｉｇｈｔｈｒｏｕｇｈｐｕｔａｎｄｈｉｇｈｅｆｆｉｃｉｅｎｃｙｓｅｐａｒａｔｉｏｎｏｆｐａｒｔｉｃｌｅｓ．Ｋｅｙｗｏｒｄｓ：ｍｉｃｒｏｆｌｕｉｄｉｃｓｃｈｉｐ；ｉｎｅｒｔｉａｌｆｏｒｃｅｓｅｐａｒａｔｉｏｎ；ｅｌｅｃｔｒｏｐｈｏｒｅｓｉｓｓｅｐａｒａｔｉｏｎ；３Ｄｅｌｅｃｔｒｏｄｅｓ０ 引 言随着微流控芯片的快速发展，与传统细胞分离技术如离心和过滤等相比，由于其样品消耗少，制备成本低，灵敏度高等潜在优势，得到了人们广泛的关注。
entropy
Entropy
Entropy —— Energy Dispersal
• 1850年，德国物理学家克劳修 1850年 斯首次提出熵的概念， 斯首次提出熵的概念，用来表 示任何一种能量在空间中分布 的均匀程度， 的均匀程度，能量分布得越均 熵就越大。 匀，熵就越大。一个体系的能 量完全均匀分布时，这个系统 量完全均匀分布时， 的熵就达到最大值。 的熵就达到最大值。 • 孤立体系，如果听任它自然发 孤立体系， 总是向着熵增大的方向。 展，总是向着熵增大的方向。 • dS=（dQ/T） dS=（dQ/T）
△r
Sm。 of the following
2C6H6（l) + 15O2 (g) → 12CO2(g) + 6H2O (l) l C6H6（l) + 15/2O2 (g) → 6CO2(g) + 3H2O (l) l) H2（g) + 1/2O2 (g) → H2O (l) 2H 2H2（g) + O2 (g) → 2 2O (l)
27.28
205.14
87.40
Principle of entropy increase
• The second law of thermodynamics
• In any real change in a isolated system the entropy always increase
戊二醛与羟基的交联机理
一种分布式用户浏览点击模型算法
第45卷第$期V ol.45 No.3计算机工程Computer Engineering2019年3月March2019•云计算与大数据专题•文章编号"1000#428(2019)0$-0001-06 文献标志码：A 中图分类号：TP391一种分布式用户浏览点击模型算法张浩盛伦1!2,李羽中柯勇张士波1(1.中国科学院计算机网络信息中心，北京100190; 2.中国科学院大学，北京100190)摘要：为从海量搜索点击日志中快速挖掘用户行为，提出一种分布式用户浏览点击模型（UBM)算法。
原始 U BM算法求出的检验度参数！只与搜索结果文档所在排序位置以及上一文档的点击位置有关，且非常稳定，基于 此特性，将E M迭代求解转换为抽样估计检验度以求解吸引度的分布式U B M算法。
在Spark数据平台上进行仿 真，结果表明，与原始U BM算法相比，该算法能够解决点击日志中存在的严重数据倾斜问题，且运行效率较高。
关键词：点击日志；点击模型；用户浏览点击模型算法；搜索引擎；Spark平台中文引用格式：张浩盛伦，李罛，柯勇，等.一种分布式用户浏览点击模型算法[J].计算机工程，2019，45(3):1-6.英文引用格式：ZHANG Haoshenglun，LI Chong，KE Yong，et al.A distributed user browse click model algorithm[J]. Computer Engineering，2019，45 (3 ): 1-6.A Distributed User Browse Click Model AlgorithmZHANG Haoshenglun1，2，LI Chong1，KE Yong1，ZHANG Shibo1(1. Computer Netw^ork Information Center，Chinese Academy of Sciences，Beijing 100190，China；2. Universit;^ of Chinese Academy of Sciences，Beijing 100190，China)[Abstract]A distributed U ser Browse Click Model#UBM)algorithm is proposed to quickly mine user behavior from massive search click logs.The validation parameter ! derived from theoriginal UBM algorithm is only related to the ranking position of the search results and the click position of the previous document，and is very stable.Based on this characteristic，the EM iteration solution i s transformed into a distributed UBM algorithm which estimates the test degree by sampling to solve the attraction degree.Results of simulation on Spark data platform show that compared with the original UBM algorithm，the proposed algorithm can solve the serious data skew problem in click log，and has higher efficiency.[Key words]click l og;click model;User Browse Click Model(UBM)algorithm;search engine;Spark platformD O I:10.19678/j.issn.1000-3428.00501190概述随着互联网技术的快速发展以及网络中海量数 据的增加，搜索引擎已经成为人们获取信息的重要 工具。
分等级大微孔、介孔大孔
ARTICLEOPENReceived11Dec2012|Accepted16May2013|Published14Jun2013A solid with a hierarchical tetramodalmicro-meso-macro pore size distributionYu Ren1,Zhen Ma2,3,Russell E.Morris1,Zheng Liu1,Feng Jiao4,Sheng Dai3&Peter G.Bruce1Porous solids have an important role in addressing some of the major energy-related pro-blems facing society.Here we describe a porous solid,a-MnO2,with a hierarchical tetramodalpore size distribution spanning the micro-,meso-and macro pore range,centred at0.48,4.0,18and70nm.The hierarchical tetramodal structure is generated by the presence ofpotassium ions in the precursor solution within the channels of the porous silica template;thesize of the potassium ion templates the microporosity of a-MnO2,whereas theirreactivity with silica leads to larger mesopores and macroporosity,without destroying themesostructure of the template.The hierarchical tetramodal pore size distribution influencesthe properties of a-MnO2as a cathode in lithium batteries and as a catalyst,changingthe behaviour,compared with its counterparts with only micropores or bimodalmicro/mesopores.The approach has been extended to the preparation of LiMn2O4with ahierarchical pore structure.1EaStCHEM,School of Chemistry,University of St Andrews,St Andrews KY169ST,UK.2Shanghai Key Laboratory of Atmospheric Particle Pollution and Prevention(LAP3),Department of Environmental Science and Engineering,Fudan University,Shanghai200433,China.3Chemical Sciences Division,Oak Ridge National Laboratory,Oak Ridge,T ennessee37831,USA.4Department of Chemical and Biomolecular Engineering,University of Delaware,Newark,Delaware19716,USA.Correspondence and requests for materials should be addressed to P.G.B.(email:p.g.bruce@).P orous solids have an important role in addressing some of the major problems facing society in the twenty-first century,such as energy storage,CO2sequestration,H2 storage,therapeutics(for example,drug delivery)and catalysis1–8. The size of the pores and their distribution directly affect their ability to function in a particular application2.For example, zeolites are used as acid catalysts in industry,but their micropores impose severe diffusion limitations on the ingress and egress of the reactants and the catalysed products9.To address such issues, great effort is being expended in preparing porous materials with a bimodal(micro and meso)pore structure by synthesizing zeolites or silicas containing micropores and mesopores10–17,or microporous metal–organic frameworks with ordered mesopores18.Among porous solids,porous transition metal oxides are particularly important,because they exhibit many unique properties due to their d-electrons and the variable redox state of their internal surfaces8,19–22.Here we describe thefirst solid(a-MnO2)possessing hierarchical pores spanning the micro,meso and macro range, centred at0.48,4.0,18and70nm.The synthesis method uses mesoporous silica as a hard template.Normally such a template generates a mesoporous solid with a unimodal23–31or,at most,a bimodal pore size distribution32–38.By incorporating Kþions in the precursor solution,within the silica template,the Kþions act bifunctionally:their size templates the formation of the micropores in a-MnO2,whereas their reactivity with silica destroys the microporous channels in KIT-6comprehensively, leading to the formation of a-MnO2containing large mesopores and,importantly,macropores,something that has not been possible by other methods.Significantly,this is achieved without destroying the silica template by alkaline ions.The effect of the tetramodal pore structure on the properties of the material is exemplified by considering their use as electrodes for lithium-ion batteries and as a catalyst for CO oxidation and N2O decomposition.The novel material offers new possibilities for combining the selectivity of small pores with the transport advantages of the large pores across a wide range of sizes.We also present results demonstrating the extension of the method to the synthesis of LiMn2O4with a hierarchical pore structure.ResultsComposition of tetramodal a-MnO2.The composition of the synthesized material was determined by atomic absorption ana-lysis and redox titration to be K0.08MnO2(the K/Mn ratio of the precursor solution was1/3).The material is commonly referred to as a-MnO2,because of the small content of Kþ19.N2sorption analysis of tetramodal a-MnO2.The tetramodal a-MnO2shows a type IV isotherm(Fig.1a).The pore size dis-tribution(Fig.1b)in the range of0.3–200nm was analysed using the density functional theory(DFT)method applied to the adsorption branch of the isotherm39–42,as this is more reliable than analysing the desorption branch43;note that this is not the DFT method used in ab initio electronic structure calculations. Plots were constructed with vertical axes representing ‘incremental pore volume’and‘incremental surface area’.Large (macro)pores can account for a significant pore volume while representing a relatively smaller surface area and vice versa for small(micro)pores.Therefore,when investigating a porous material with a wide range of pore sizes,for example,micropore and macropore,the combination of surface area and pore volume is essential to determine the pore size distribution satisfactorily (Fig.1b).Considering both pore volume and surface area, significant proportions of micro-,meso-and macropores are evident,with distinct maxima centred at0.70,4.0,18and70nm.To probe the size of the micropores more precisely than is possible with DFT,the Horvath–Kawazoe pore size distribution analysis was employed44.A single peak was obtained at0.48nm(Fig.1c),in good accord with the0.46-nm size of the2Â2channels of a-MnO2 (refs.19,21).The relatively small Brunauer–Emmett–Teller(BET) surface area of tetramodal a-MnO2(79–105m2gÀ1; Supplementary Table S1)compared with typical surface areas of mesoporous metal oxides(90–150m2gÀ1)45is due to the significant proportion of macropores(which have small surface areas)and relatively large(18nm)mesopores—a typical mesoporous metal oxide has only3–4nm pores.TEM analysis of tetramodal a-MnO2.Transmission electron microscopic(TEM)data for tetramodal a-MnO2,Fig.2, demonstrates a three-dimensional pore structure with a sym-metry consistent with space group Ia3d.From the TEM data,an a0lattice parameter of23.0nm for the mesostructure could be extracted,which is in good agreement with the value obtained from the low-angle powder X-ray diffraction(PXRD)data, a0¼23.4nm(Supplementary Fig.S1a).High-resolution TEM images in Fig.2c–e demonstrate that the walls are crystalline with a typical wall thickness of10nm.The lattice spacings of0.69,0.31 and0.35nm agree well with the values of6.92,3.09and3.46Åfor the[110],[310]and[220]planes of a-MnO2(International Centre for Diffraction Data(ICDD)number00-044-0141), respectively.The wide-angle PXRD data matches well with the PXRD data of bulk cryptomelane a-MnO2(Supplementary Fig. S1b),confirming the crystalline walls.The various pores in tetramodal a-MnO2can be observed by TEM directly:the0.48-nm micropores are seen in Fig.2e(2Â2 tunnels with dimensions of0.48Â0.48nm in the white box);the 4.0-nm pores are shown in Fig.2b–d;the18-nm pores are shown in Fig.2a;the70-nm pores are evident in Fig.2b(highlighted with white circles).Li intercalation.Li can be intercalated into bulk a-MnO2 (ref.46).Therefore,it is interesting to compare Li intercalation into bulk a-MnO2(micropores only)and bimodal a-MnO2 (micropores along with a single mesopore of diameter3.6nm,see Methods)with tetramodal a-MnO2(micro-,meso-and macropores).Each of the three a-MnO2materials was subjected to Li intercalation by incorporation as the positive electrode in a lithium battery,along with a lithium anode and a non-aqueous electrolyte(see Methods).The results of cycling(repeated intercalation/deintercalation of Li)the cells are shown in Fig.3. Although all exhibit good capacity to cycle Li at low rates of charge/discharge(30mA gÀ1),tetramodal a-MnO2shows sig-nificantly higher capacity(Li storage)at a high rate of 6,000mA gÀ1(corresponding to charge and discharge in3min). The tetramodal a-MnO2can store three times the capacity(Li) compared with bimodal a-MnO2,and18times that of a-MnO2 with only micropores,at the high rate of intercalation/deinter-calation(Fig.3).The superior rate capability of tetramodal a-MnO2over microporous and bimodal forms may be assigned to better Liþtransport in the electrolyte within the hierarchical pore structure of tetramodal a-MnO2.The importance of elec-trolyte transport in porous electrodes has been discussed recently35,47,48and the results presented here reinforce the beneficial effect of a hierarchical pore structure.Catalytic studies.CO oxidation and N2O decomposition were used as reactions to probe the three different forms of a-MnO2as catalysts(Supplementary Fig.S2).As shown in Supplementary Fig.S2a,tetramodal a-MnO2demonstrates better catalytic activity compared with only micropores or bimodal a-MnO2;thetemperature of half CO conversion (T 50)was 124°C for tetra-modal a -MnO 2,whereas microporous and bimodal a -MnO 2exhibited a T 50value of 275°C and 209°C,respectively.In the case of N 2O decomposition,a -MnO 2with only micropores demonstrated no catalytic activity in the range of 200–400°C,in accord with a previous report 49.Tetramodal and bimodal a -MnO 2showed catalytic activity and reached 32%and 20%of N 2O conversion,respectively,at a reaction temperature of 400°C.The differences in catalytic activity are related to the differences in the material.A detailed study focusing on the catalytic activity alonewould be necessary to demonstrate which specific features of the textural differences (pore size distribution,average manganese oxidation state,K þand so on)between the different MnO 2materials are responsible for the differences in behaviour.However,the preliminary results shown here do illustrate that such differences exist.Porous LiMn 2O 4.To demonstrate the wider applicability of the synthesis method,LiMn 2O 4with a hierarchical pore structurewas1801601401201008060402000.00.20.40.60.81.0V (c m 3 g –1)Pore diameter (nm)0.0120.0100.0080.0060.0040.0020.000I n c r e m e n t a l p o r e v o l u m e (c m 3 g –1)Pore width (nm)I n c r e m e n t a l s u r f a c e a r e a (m 2 g –1)I n c r e m e n t a l s u r f a c e a r e a (m 2 g –1)P /P 0Figure 1|N 2sorption analysis of tetramodal a -MnO 2.(a )N 2adsorption–desorption isotherms,(b )DFT pore size distribution and (c )Horvath–Kawazoe pore size distribution from N 2adsorption isotherm for tetramodal a -MnO 2.Figure 2|TEM images of tetramodal a -MnO 2.TEM images along (a )[100]direction,showing 18nm mesopores (scale bar,50nm);(b )4.0and 70nm pores (70nm pores are highlighted by white circles;scale bar,100nm);(c –e )high-resolution (HRTEM)images of tetramodal a -MnO 2showing 4.0and 0.48nm pores (scale bar,10nm).Inset is representation of a -MnO 2structure along the c axis,demonstrating the 2Â2micropores as shown in the HRTEM (white box)in e .Purple,octahedral MnO 6;red,oxygen;violet,potassium.synthesized in a way similar to that of tetramodal a -MnO 2.The main difference is the use of LiNO 3instead of KNO 3(see Methods).In this case,Li þreacts with the silica template col-lapsing/blocking the microporous channels in the KIT-6and resulting in the large mesopores and macropores (17and 50nm)in the LiMn 2O 4obtained.The use of Li þinstead of the larger K þdeters the formation of micropores because Li þis too small.TEM analysis illustrates the hierarchical pore structure of LiMn 2O 4(Supplementary Fig.S3):4.0nm pores are evident in Supplementary Fig.S3b;17nm pores in Supplementary Fig.S3a;and 50nm pores in Supplementary Fig.S3b (highlighted with white circles).The d-spacing of 0.47nm in the high-resolution TEM image (Supplementary Fig.S3c)is in good accordance with the values of 0.4655nm for the [111]planes of LiMn 2O 4(ICDD number 00-038-0789)and with the wide-angle PXRD data (Supplementary Fig.S4).The original DFT pore size distribution analysis from N 2sorption (adsorption branch)gives three pore sizes in the range of 1–100nm centred at 4.0,17and 50nm (Supplementary Fig.S5).A more in-depth presentation of the results for LiMn 2O 4will be given in a future paper;preliminary results presented here illustrate that the basic method can be applied beyond a -MnO 2.DiscussionTurning to the synthesis of the tetramodal a -MnO 2,the details are given in the Methods section.Hard templating using silica templates,such as KIT-6,normally gives rise to materials with unimodal or,at most,bimodal mesopore structures,and in the latter case the smaller mesopores dominate over the larger mesopores 8,32,35.Alkali ions are excellent templates for micropores in transition metal oxides 19,21,but they have been avoided in nanocasting from silica templates because of concerns that they would react with and,hence,destroy thesilica20018016014012010080604020D i s c h a r g e c a p a c i t y (m A h g –1)0Cycle numberx in Li x MnO 2Figure 3|Electrochemical behaviour of different a -MnO 2.Capacity retention for tetramodal a -MnO 2cycled at 30(empty blue circles)and 6,000mA g À1(filled blue circles);bulk a -MnO 2cycled at 30(empty red squares)and 6,000mA g À1(filled red squares);bimodal a -MnO 2cycled at 30(empty black triangle)and 6,000mA g À1(filled blacktriangles).18 nm pores70 nm poresTwo sets of mesoporeschannels connecting both sets of mesoporesEtching of silica Etching of silica Etching of silica template2discontinuously within one set of the KIT-6mesoporesFigure 4|Formation mechanism of meso and macropores in tetramodal a -MnO 2.When both KIT-6mesochannels are occupied by a -MnO 2and then the silica between them etched away,the remaining pore is 4nm (centre portion of figure).When a -MnO 2grows in only one set of mesochannels and then the KIT-6is dissolved away,the remaining metal oxide has 18nm pores (upper portion of figure).The comprehensive destruction of the microchannels in KIT-6by K þleads to a -MnO 2growing in only a proportion of one set of the KIT-6mesochannels,resulting in the formation of B 70nm pores (lower portion of figure).template50.Here,not only have alkali ions been used successfully in precursor solutions without destroying the template mesostructure but they give rise to macropores in the a-MnO2, thus permitting the synthesis of a tetramodal,micro-small,meso-large,meso-macro pore structure.Synthesis begins by impregnating the KIT-6silica template with a precursor solution containing Mn2þand Kþions.On heating,the Kþions template the formation of the micropores in a-MnO2,as the latter forms within the KIT-6template.KIT-6 consists of two interpenetrating mesoporous channels linked by microporous channels51–53.The branches of the two different sets of mesoporous channels in KIT-6are nearest neighbours separated by a silica wall of B4nm53;therefore,when both KIT-6mesochannels are occupied by a-MnO2and the silica between them etched away,the remaining pore is4nm(see centre portion of Fig.4).It has been shown previously,by a number of authors,that by varying the hydrothermal conditions used to prepare the KIT-6,the proportion of the microchannels can be decreased to some extent,thus making it difficult to simultaneouslyfill the neighbouring KIT-6mesoporous channels by the precursor solution of the target mesoporous metal oxide33–35.As a result,the target metal oxide grows in only one set of mesochannels of the KIT-6host but not both.When the KIT-6is dissolved away,the remaining metal oxide has B18nm pores,because the distance between adjacent branches of the same KIT-6mesochannels is greater than between the two different mesochannels in KIT-6.Here we propose that the Kþions have a similar effect on the KIT-6to that of the hydrothermal synthesis,but by a completely different mechanism.Reaction between the Kþions in the precursor solution with the silica during calcination results in the formation of Kþ-silicates,which cause collapse or blocking of the microporous channels in KIT-6,such that the a-MnO2grows in one set of the KIT-6mesochannels,giving rise to18nm pores in a-MnO2when the silica is etched away,see top portion of Fig.4. However,the reaction between Kþand the silica is more severe than the effect of varying the hydrothermal treatment.In the former case,the KIT-6microchannels are so comprehensively destroyed that the proportion of the large(18nm)to smaller (4nm)mesopores is greater than can be achieved by varying hydrothermal conditions.The comprehensive destruction of the microchannels in KIT-6by Kþ,perhaps augmented by some minor degradation of parts of the mesochannels,leads to a-MnO2 growing in only a proportion of one set of the KIT-6 mesochannels,resulting in the formation of B70nm pores in a-MnO2,see lower portion of Fig.4.In summary,the Kþreactivity with the silica goes beyond what can be achieved by varying the conditions of hydrothermal synthesis and is responsible for generating the tetramodal pore size distribution reported here. The mechanism of pore formation in a-MnO2by reaction between Kþand the silica template is supported by several findings.First,by the lower K/Mn molar ratio of thefinal tetramodal a-MnO2product(0.08)compared with the starting materials(0.33)implies that some of the Kþions in the impregnating solution have reacted with the silica.Second, support for collapse/blocking of the microporous channels in KIT-6due to reaction with Kþwas obtained by comparing the texture of KIT-6impregnated with an aqueous solution contain-ing only KNO3and calcined at300and500°C.The micropore volume in KIT-6is the greatest,with no KNO3in the solution;it then decreases continuously as the calcination temperature and calcination time is increased,such that after2and5h at500°C the micropore volume has decreased to zero(Supplementary Fig. S6).Third,we prepared tetramodal a-MnO2using a similar synthetic procedure to that described in the Methods section, except that this time we used a covered tall crucible for the calcination step.Sun et al.54have shown that using a covered,tall crucible when calcining results in porous metal oxides with much larger particle sizes.If the70-nm pores had arisen simply from the gaps between the particles,then the pore size would have changed;in contrast,it remained centred at70nm, Supplementary Fig.S7,consistent with the70-nm pores being intrinsic to the materials and arising from reaction with the Kþas described above.Fourth,if the synthesis of MnO2is carried out using the KIT-6template but in the absence Kþions,then the DFT pore size distribution shown in Supplementary Fig.S8is obtained.The0.48-and70-nm pores are now absent,but the4-and18-nm pores remain.This demonstrates the key role of Kþin the formation of the smallest and largest pores and,hence,in generating the tetramodal pore size distribution.The absence of Kþmeans that there is nothing to template the0.48nm pores and so a-MnO2is not formed;the b-polymorph is obtained instead.The absence of Kþalso means that the microchannels in the KIT-6template remain intact,resulting in no70nm pores and the dominance of the4-nm pores compared with the 18-nm pores.The hierarchical pore structure can be varied systematically by controlling the synthesis conditions,in particular the Kþ/Mn ratio of the precursor solution.A range of Kþ/Mn ratios,1/5,1/3and1/2,gave rise to a series of pore size distributions,in which the pore sizes remained the same but the relative proportions of the different pores varied (Supplementary Table S1).The higher the Kþ/Mn ratio,the greater the proportion of macropores and large mesopores.This is in accord with expectations,as the higher the Kþconcentra-tion in the precursor solution the greater the collapse/blocking of the microporous channels in the KIT-6(as noted above),and hence the greater the proportion of macropores and large mesopores.Indeed,these results offer further support for the mechanism of pore size distribution arising from reaction between Kþand the silica template.In conclusion,tetramodal a-MnO2,thefirst porous solid with a tetramodal pore size distribution,has been synthesized.Its hierarchical pore structure spans the micro,meso and macropore range between0.3and200nm,with pore dimensions centred at 0.48,4.0,18and70nm.Key to the synthesis is the use of Kþions that not only template the formation of micropores but also react with the silica template,therefore,breaking/blocking the micro-porous channels in the silica template far more comprehensively than is possible by varying the hydrothermal synthesis conditions, to the extent that macropores are formed,and without destroying the silica mesostructure by alkali ions,as might have been expected.The resulting hierarchical tetramodal structure demon-strates different behaviours compared with microporous and bimodal a-MnO2as a cathode material for Li-ion batteries,and when used as a catalyst for CO oxidation and N2O decomposi-tion.The method has been extended successfully to the preparation of hierarchical LiMn2O4.MethodsSynthesis.Tetramodal a-MnO2(surface area96m2gÀ1,K0.08MnO2)was pre-pared by two-solvent impregnation55using Kþand mesoporous silica KIT-6as the hard template.KIT-6was prepared according to a previous report (hydrothermal treatment at100°C)51.In a typical synthesis of tetramodal a-MnO2, 7.53g of Mn(NO3)2Á4H2O(98%,Aldrich)and1.01g of KNO3(99%,Aldrich)were dissolved in B10ml of water to form a solution with a molar ratio of Mn/K¼3.0. Next,5g of KIT-6was dispersed in200ml of n-hexane.After stirring at room temperature for3h,5ml of the Mn/K solution was added slowly with stirring.The mixture was stirred overnight,filtered and dried at room temperature until a completely dried powder was obtained.The sample was heated slowly to500°C (1°C minÀ1),calcined at that temperature for5h with a cover in a normal crucible unless is specified54and the resulting material treated three times with a hot aqueous KOH solution(2.0M),to remove the silica template,followed by washing with water and ethanol several times,and then drying at60°C.Bimodal a-MnO2(surface area58m2gÀ1,K0.06MnO2)with micropore and a single mesopore size of3.6nm was prepared by using mesoporous silica SBA-15as a hard template.The SBA-15was prepared according to a previous report56.Bulk a-MnO2(surface area8m2gÀ1,K0MnO2)was prepared by the reaction between325mesh Mn2O3(99.0%,Aldrich)and6.0M H2SO4solution at80°C for 24h,resulting in the disproportionation of Mn2O3into a soluble Mn2þspecies and the desired a-MnO2product46.Treatment of KIT-6with KNO3was carried out as follows:1.01g of KNO3was dissolved in B15ml of water to form a KNO3solution.Five grams of mesoporous KIT-6was dispersed in200ml of n-hexane.After stirring at room temperature for 3h,5ml of KNO3solution was added slowly with stirring.The mixture was stirred overnight,filtered and dried at room temperature until a completely dried powder was obtained.The sample was heated slowly to300or500°C(1°C minÀ1), calcined at that temperature for5h and the resulting material was washed with water and ethanol several times,and then dried at60°C overnight.The synthesis method for hierarchical porous LiMn2O4was similar to that of tetramodal a-MnO2.The main difference was to use1.01g of LiNO3instead of KNO3.After impregnation into KIT-6,calcination and silica etching,porous LiMn2O4was obtained.Characterization.TEM studies were carried out using a JEOL JEM-2011, employing a LaB6filament as the electron source,and an accelerating voltage of 200keV.TEM images were recorded by a Gatan charge-coupled device camera in a digital format.Wide-angle PXRD data were collected on a Stoe STADI/P powder diffractometer operating in transmission mode with Fe K a1source radiation(l¼1.936Å).Low-angle PXRD data were collected using a Rigaku/MSC,D/max-rB with Cu K a1radiation(l¼1.541Å)operating in reflection mode with a scintillation detector.N2adsorption–desorption analysis was carried out using a Micromeritics ASAP2020.The typical sample weight used was100–200mg. The outgas condition was set to300°C under vacuum for2h,and all adsorption–desorption measurements were carried out at liquid nitrogen tem-perature(À196°C).The original DFT method for the slit pore geometry was used to extract the pore size distribution from the adsorption branch usingthe Micromeritics software39–42.A Horvath–Kawazoe method was used to extract the microporosity44.Mn and K contents were determined by chemical analysis using a Philips PU9400X atomic adsorption spectrometer.The average oxidation state of framework manganese in a-MnO2samples was determined by a redoxtitration method57.Electrochemistry.First,the cathode was constructed by mixing the active material (a-MnO2),Kynar2801(a copolymer based on polyvinylidenefluoride),and Super S carbon(MMM)in the weight ratio80:10:10.The mixture was cast onto Al foil (99.5%,thickness0.050mm,Advent Research Materials,Ltd)from acetone using a Doctor-Blade technique.After solvent evaporation at room temperature and heating at80°C under vacuum for8h,the cathode was assembled into cells along with a Li metal anode and electrolyte(Merck LP30,1M LiPF6in1:1v/v ethylene carbonate/dimethyl carbonate).The cells were constructed and handled in anAr-filled MBraun glovebox(O2o0.1p.p.m.,H2O o0.1p.p.m.).Electrochemical measurements were carried out at30°C using a MACCOR Series4200cycler.Catalysis.Catalytic CO oxidation was tested in a plug-flow microreactor(Alta-mira AMI200).Fifty milligrams of catalyst was loaded into a U-shaped quartz tube (4mm i.d.).After the catalyst was pretreated inflowing8%O2(balanced with He) at400°C for1h,the catalyst was then cooled down,the gas stream switched to1% CO(balanced with air)and the reaction temperature ramped using a furnace(at a rate of1°C minÀ1above ambient temperature)to record the light-off curve.The flow rate of the reactant stream was37cm3minÀ1.A portion of the product stream was extracted periodically with an automatic sampling valve and was analysed using a dual column gas chromatograph with a thermal conductivity detector.To perform N2O decomposition reaction testing,0.5g catalyst was packed into a U-shaped glass tube(7mm i.d.)sealed by quartz wool,and pretreated inflowing 20%O2(balance He)at400°C for1h(flow rate:50cm3minÀ1).After cooling to near-room temperature,a gas stream of0.5%N2O(balance He)flowed through the catalyst at a rate of60cm3minÀ1,and the existing stream was analysed by a gas chromatograph(Agilent7890A)that separates N2O,O2and N2.The reaction temperature was varied using a furnace,and kept at100,150,200,250,300,350 and400°C for30min at each reaction temperature.The N2O conversion determined from GC analysis was denoted as X¼([N2O]in—[N2O]out)/[N2O]inÂ100%.References1.Corma,A.From microporous to mesoporous molecular sieve materials andtheir use in catalysis.Chem.Rev.97,2373–2419(1997).2.Davis,M.E.Ordered porous materials for emerging applications.Nature417,813–821(2002).3.Taguchi,A.&Schu¨th,F.Ordered mesoporous materials in catalysis.Micro.Meso.Mater.77,1–45(2005).4.Fe´rey,G.Hybrid porous solids:past,present,future.Chem.Soc.Rev.37,191–214(2008).5.Bruce,P.G.,Scrosati,B.&Tarascon,J.M.Nanomaterials for rechargeablelithium batteries.Angew.Chem.Int.Ed.47,2930–2946(2008).6.Zhai,Y.et al.Carbon materials for chemical capacitive energy storage.Adv.Mater.23,4828–4850(2011).7.Tu¨ysu¨z,H.&Schu¨th,F.in Advances in Catalysis.Chapter Two Vol.55pp127–239(Academic Press,2012).8.Ren,Y.,Ma,Z.&Bruce,P.G.Ordered mesoporous metal oxides:synthesis andapplications.Chem.Soc.Rev.41,4909–4927(2012).9.Corma,A.State of the art and future challenges of zeolites as catalysts.J.Catal.216,298–312(2003).10.Liu,Y.,Zhang,W.&Pinnavaia,T.J.Steam-stable aluminosilicatemesostructures assembled from zeolite type Y seeds.J.Am.Chem.Soc.122, 8791–8792(2000).11.Meng,X.J.,Nawaz,F.&Xiao,F.S.Templating route for synthesizingmesoporous zeolites with improved catalytic properties.Nano Today4,292–301(2009).12.Lopez-Orozco,S.,Inayat,A.,Schwab,A.,Selvam,T.&Schwieger,W.Zeoliticmaterials with hierarchical porous structures.Adv.Mater.23,2602–2615(2011).13.Na,K.et al.Directing zeolite structures into hierarchically nanoporousarchitectures.Science333,328–332(2011).14.Chen,L.-H.et al.Hierarchically structured zeolites:synthesis,mass transportproperties and applications.J.Mater.Chem.22,17381–17403(2012).15.Tsapatsis,M.Toward high-throughput zeolite membranes.Science334,767–768(2011).16.Zhang,X.et al.Synthesis of self-pillared zeolite nanosheets by repetitivebranching.Science336,1684–1687(2012).17.Jiang,J.et al.Synthesis and structure determination of the hierarchicalmeso-microporous zeolite ITQ-43.Science333,1131–1134(2011).18.Zhao,Y.et al.Metal–organic framework nanospheres with well-orderedmesopores synthesized in an ionic liquid/CO2/surfactant system.Angew.Chem.Int.Ed.50,636–639(2011).19.Feng,Q.,Kanoh,H.&Ooi,K.Manganese oxide porous crystals.J.Mater.Chem.9,319–333(1999).20.Tiemann,M.Repeated templating.Chem.Mater.20,961–971(2008).21.Suib,S.L.Structure,porosity,and redox in porous manganese oxide octahedrallayer and molecular sieve materials.J.Mater.Chem.18,1623–1631(2008).22.Zheng,H.et al.Nanostructured tungsten oxide–properties,synthesis,andapplications.Adv.Funct.Mater.21,2175–2196(2011).ha,S.C.&Ryoo,R.Synthesis of thermally stable mesoporous cerium oxidewith nanocrystalline frameworks using mesoporous silica templates.Chem.Commun.39,2138–2139(2003).24.Tian,B.Z.et al.General synthesis of ordered crystallized metal oxidenanoarrays replicated by microwave-digested mesoporous silica.Adv.Mater.15,1370–1374(2003).25.Zhu,K.K.,Yue,B.,Zhou,W.Z.&He,H.Y.Preparation of three-dimensionalchromium oxide porous single crystals templated by mun.39,98–99(2003).26.Tian,B.Z.et al.Facile synthesis and characterization of novel mesoporous andmesorelief oxides with gyroidal structures.J.Am.Chem.Soc.126,865–875 (2004).27.Jiao,F.,Shaju,K.M.&Bruce,P.G.Synthesis of nanowire and mesoporouslow-temperature LiCoO2by a post-templating reaction.Angew.Chem.Int.Ed.44,6550–6553(2005).28.Rossinyol,E.et al.Nanostructured metal oxides synthesized by hard templatemethod for gas sensing applications.Sens.Actuator B Chem.109,57–63(2005).29.Shen,W.H.,Dong,X.P.,Zhu,Y.F.,Chen,H.R.&Shi,J.L.MesoporousCeO2and CuO-loaded mesoporous CeO2:Synthesis,characterization,and CO catalytic oxidation property.Micro.Meso.Mater.85,157–162(2005).30.Wang,Y.Q.et al.Weakly ferromagnetic ordered mesoporous Co3O4synthesized by nanocasting from vinyl-functionalized cubic Ia3d mesoporous silica.Adv.Mater.17,53–56(2005).31.Ren,Y.et al.Ordered crystalline mesoporous oxides as catalysts for COoxidation.Catal.Lett.131,146–154(2009).32.Jiao,K.et al.Growth of porous single-crystal Cr2O3in a3-D mesopore system.mun.41,5618–5620(2005).33.Rumplecker,A.,Kleitz,F.,Salabas,E.L.&Schu¨th,F.Hard templating pathwaysfor the synthesis of nanostructured porous Co3O4.Chem.Mater.19,485–496 (2007).34.Jiao,F.et al.Synthesis of ordered mesoporous NiO with crystalline walls anda bimodal pore size distribution.J.Am.Chem.Soc.130,5262–5266(2008).35.Ren,Y.,Armstrong,A.R.,Jiao,F.&Bruce,P.G.Influence of size on therate of mesoporous electrodes for lithium batteries.J.Am.Chem.Soc.132, 996–1004(2010).。
微生物英文文献及翻译—原文
Dynamic and distribution of ammonia-oxidizing bacteria communities during sludge granulation in an anaerobic e aerobic sequencing batch reactorZhang Bin a ,b ,Chen Zhe a ,b ,Qiu Zhigang a ,b ,Jin Min a ,b ,Chen Zhiqiang a ,b ,Chen Zhaoli a ,b ,Li Junwen a ,b ,Wang Xuan c ,*,Wang Jingfeng a ,b ,**aInstitute of Hygiene and Environmental Medicine,Academy of Military Medical Sciences,Tianjin 300050,PR China bTianjin Key Laboratory of Risk Assessment and Control for Environment and Food Safety,Tianjin 300050,PR China cTianjin Key Laboratory of Hollow Fiber Membrane Material and Membrane Process,Institute of Biological and Chemical Engineering,Tianjin Polytechnical University,Tianjin 300160,PR Chinaa r t i c l e i n f oArticle history:Received 30June 2011Received in revised form 10September 2011Accepted 10September 2011Available online xxx Keywords:Ammonia-oxidizing bacteria Granular sludgeCommunity development Granule sizeNitrifying bacteria distribution Phylogenetic diversitya b s t r a c tThe structure dynamic of ammonia-oxidizing bacteria (AOB)community and the distribution of AOB and nitrite-oxidizing bacteria (NOB)in granular sludge from an anaerobic e aerobic sequencing batch reactor (SBR)were investigated.A combination of process studies,molecular biotechniques and microscale techniques were employed to identify and characterize these organisms.The AOB community structure in granules was substantially different from that of the initial pattern of the inoculants sludge.Along with granules formation,the AOB diversity declined due to the selection pressure imposed by process conditions.Denaturing gradient gel electrophoresis (DGGE)and sequencing results demonstrated that most of Nitrosomonas in the inoculating sludge were remained because of their ability to rapidly adapt to the settling e washing out action.Furthermore,DGGE analysis revealed that larger granules benefit more AOB species surviving in the reactor.In the SBR were various size granules coexisted,granule diameter affected the distribution range of AOB and NOB.Small and medium granules (d <0.6mm)cannot restrict oxygen mass transfer in all spaces of the rger granules (d >0.9mm)can result in smaller aerobic volume fraction and inhibition of NOB growth.All these observations provide support to future studies on the mechanisms responsible for the AOB in granules systems.ª2011Elsevier Ltd.All rights reserved.1.IntroductionAt sufficiently high levels,ammonia in aquatic environments can be toxic to aquatic life and can contribute to eutrophica-tion.Accordingly,biodegradation and elimination of ammonia in wastewater are the primary functions of thewastewater treatment process.Nitrification,the conversion of ammonia to nitrate via nitrite,is an important way to remove ammonia nitrogen.It is a two-step process catalyzed by ammonia-oxidizing and nitrite-oxidizing bacteria (AOB and NOB).Aerobic ammonia-oxidation is often the first,rate-limiting step of nitrification;however,it is essential for the*Corresponding author .**Corresponding author.Institute of Hygiene and Environmental Medicine,Academy of Military Medical Sciences,Tianjin 300050,PR China.Tel.:+862284655498;fax:+862223328809.E-mail addresses:wangxuan0116@ (W.Xuan),jingfengwang@ (W.Jingfeng).Available online atjournal homepage:/locate/watresw a t e r r e s e a r c h x x x (2011)1e 100043-1354/$e see front matter ª2011Elsevier Ltd.All rights reserved.doi:10.1016/j.watres.2011.09.026removal of ammonia from the wastewater(Prosser and Nicol, 2008).Comparative analyses of16S rRNA sequences have revealed that most AOB in activated sludge are phylogeneti-cally closely related to the clade of b-Proteobacteria (Kowalchuk and Stephen,2001).However,a number of studies have suggested that there are physiological and ecological differences between different AOB genera and lineages,and that environmental factors such as process parameter,dis-solved oxygen,salinity,pH,and concentrations of free ammonia can impact certain species of AOB(Erguder et al., 2008;Kim et al.,2006;Koops and Pommerening-Ro¨ser,2001; Kowalchuk and Stephen,2001;Shi et al.,2010).Therefore, the physiological activity and abundance of AOB in waste-water processing is critical in the design and operation of waste treatment systems.For this reason,a better under-standing of the ecology and microbiology of AOB in waste-water treatment systems is necessary to enhance treatment performance.Recently,several developed techniques have served as valuable tools for the characterization of microbial diversity in biological wastewater treatment systems(Li et al., 2008;Yin and Xu,2009).Currently,the application of molec-ular biotechniques can provide clarification of the ammonia-oxidizing community in detail(Haseborg et al.,2010;Tawan et al.,2005;Vlaeminck et al.,2010).In recent years,the aerobic granular sludge process has become an attractive alternative to conventional processes for wastewater treatment mainly due to its cell immobilization strategy(de Bruin et al.,2004;Liu et al.,2009;Schwarzenbeck et al.,2005;Schwarzenbeck et al.,2004a,b;Xavier et al.,2007). Granules have a more tightly compact structure(Li et al.,2008; Liu and Tay,2008;Wang et al.,2004)and rapid settling velocity (Kong et al.,2009;Lemaire et al.,2008).Therefore,granular sludge systems have a higher mixed liquid suspended sludge (MLSS)concentration and longer solid retention times(SRT) than conventional activated sludge systems.Longer SRT can provide enough time for the growth of organisms that require a long generation time(e.g.,AOB).Some studies have indicated that nitrifying granules can be cultivated with ammonia-rich inorganic wastewater and the diameter of granules was small (Shi et al.,2010;Tsuneda et al.,2003).Other researchers reported that larger granules have been developed with the synthetic organic wastewater in sequencing batch reactors(SBRs)(Li et al., 2008;Liu and Tay,2008).The diverse populations of microor-ganisms that coexist in granules remove the chemical oxygen demand(COD),nitrogen and phosphate(de Kreuk et al.,2005). However,for larger granules with a particle diameter greater than0.6mm,an outer aerobic shell and an inner anaerobic zone coexist because of restricted oxygen diffusion to the granule core.These properties of granular sludge suggest that the inner environment of granules is unfavorable to AOB growth.Some research has shown that particle size and density induced the different distribution and dominance of AOB,NOB and anam-mox(Winkler et al.,2011b).Although a number of studies have been conducted to assess the ecology and microbiology of AOB in wastewater treatment systems,the information on the dynamics,distribution,and quantification of AOB communities during sludge granulation is still limited up to now.To address these concerns,the main objective of the present work was to investigate the population dynamics of AOB communities during the development of seedingflocs into granules,and the distribution of AOB and NOB in different size granules from an anaerobic e aerobic SBR.A combination of process studies,molecular biotechniques and microscale techniques were employed to identify and char-acterize these organisms.Based on these approaches,we demonstrate the differences in both AOB community evolu-tion and composition of theflocs and granules co-existing in the SBR and further elucidate the relationship between distribution of nitrifying bacteria and granule size.It is ex-pected that the work would be useful to better understand the mechanisms responsible for the AOB in granules and apply them for optimal control and management strategies of granulation systems.2.Material and methods2.1.Reactor set-up and operationThe granules were cultivated in a lab-scale SBR with an effective volume of4L.The effective diameter and height of the reactor was10cm and51cm,respectively.The hydraulic retention time was set at8h.Activated sludge from a full-scale sewage treat-ment plant(Jizhuangzi Sewage Treatment Works,Tianjin, China)was used as the seed sludge for the reactor at an initial sludge concentration of3876mg LÀ1in MLSS.The reactor was operated on6-h cycles,consisting of2-min influent feeding,90-min anaerobic phase(mixing),240-min aeration phase and5-min effluent discharge periods.The sludge settling time was reduced gradually from10to5min after80SBR cycles in20days, and only particles with a settling velocity higher than4.5m hÀ1 were retained in the reactor.The composition of the influent media were NaAc(450mg LÀ1),NH4Cl(100mg LÀ1),(NH4)2SO4 (10mg LÀ1),KH2PO4(20mg LÀ1),MgSO4$7H2O(50mg LÀ1),KCl (20mg LÀ1),CaCl2(20mg LÀ1),FeSO4$7H2O(1mg LÀ1),pH7.0e7.5, and0.1mL LÀ1trace element solution(Li et al.,2007).Analytical methods-The total organic carbon(TOC),NHþ4e N, NOÀ2e N,NOÀ3e N,total nitrogen(TN),total phosphate(TP) concentration,mixed liquid suspended solids(MLSS) concentration,and sludge volume index at10min(SVI10)were measured regularly according to the standard methods (APHA-AWWA-WEF,2005).Sludge size distribution was determined by the sieving method(Laguna et al.,1999).Screening was performed with four stainless steel sieves of5cm diameter having respective mesh openings of0.9,0.6,0.45,and0.2mm.A100mL volume of sludge from the reactor was sampled with a calibrated cylinder and then deposited on the0.9mm mesh sieve.The sample was subsequently washed with distilled water and particles less than0.9mm in diameter passed through this sieve to the sieves with smaller openings.The washing procedure was repeated several times to separate the gran-ules.The granules collected on the different screens were recovered by backwashing with distilled water.Each fraction was collected in a different beaker andfiltered on quantitative filter paper to determine the total suspended solid(TSS).Once the amount of total suspended solid(TSS)retained on each sieve was acquired,it was reasonable to determine for each class of size(<0.2,[0.2e0.45],[0.45e0.6],[0.6e0.9],>0.9mm) the percentage of the total weight that they represent.w a t e r r e s e a r c h x x x(2011)1e10 22.2.DNA extraction and nested PCR e DGGEThe sludge from approximately8mg of MLSS was transferred into a1.5-mL Eppendorf tube and then centrifuged at14,000g for10min.The supernatant was removed,and the pellet was added to1mL of sodium phosphate buffer solution and aseptically mixed with a sterilized pestle in order to detach granules.Genomic DNA was extracted from the pellets using E.Z.N.A.äSoil DNA kit(D5625-01,Omega Bio-tek Inc.,USA).To amplify ammonia-oxidizer specific16S rRNA for dena-turing gradient gel electrophoresis(DGGE),a nested PCR approach was performed as described previously(Zhang et al., 2010).30m l of nested PCR amplicons(with5m l6Âloading buffer)were loaded and separated by DGGE on polyacrylamide gels(8%,37.5:1acrylamide e bisacrylamide)with a linear gradient of35%e55%denaturant(100%denaturant¼7M urea plus40%formamide).The gel was run for6.5h at140V in 1ÂTAE buffer(40mM Tris-acetate,20mM sodium acetate, 1mM Na2EDTA,pH7.4)maintained at60 C(DCodeäUniversal Mutation Detection System,Bio-Rad,Hercules,CA, USA).After electrophoresis,silver-staining and development of the gels were performed as described by Sanguinetti et al. (1994).These were followed by air-drying and scanning with a gel imaging analysis system(Image Quant350,GE Inc.,USA). The gel images were analyzed with the software Quantity One,version4.31(Bio-rad).Dice index(Cs)of pair wise community similarity was calculated to evaluate the similarity of the AOB community among DGGE lanes(LaPara et al.,2002).This index ranges from0%(no common band)to100%(identical band patterns) with the assistance of Quantity One.The Shannon diversity index(H)was used to measure the microbial diversity that takes into account the richness and proportion of each species in a population.H was calculatedusing the following equation:H¼ÀPn iNlogn iN,where n i/Nis the proportion of community made up by species i(bright-ness of the band i/total brightness of all bands in the lane).Dendrograms relating band pattern similarities were automatically calculated without band weighting(consider-ation of band density)by the unweighted pair group method with arithmetic mean(UPGMA)algorithms in the Quantity One software.Prominent DGGE bands were excised and dissolved in30m L Milli-Q water overnight,at4 C.DNA was recovered from the gel by freeze e thawing thrice.Cloning and sequencing of the target DNA fragments were conducted following the estab-lished method(Zhang et al.,2010).2.3.Distribution of nitrifying bacteriaThree classes of size([0.2e0.45],[0.45e0.6],>0.9mm)were chosen on day180for FISH analysis in order to investigate the spatial distribution characteristics of AOB and NOB in granules.2mg sludge samples werefixed in4%para-formaldehyde solution for16e24h at4 C and then washed twice with sodium phosphate buffer;the samples were dehydrated in50%,80%and100%ethanol for10min each. Ethanol in the granules was then completely replaced by xylene by serial immersion in ethanol-xylene solutions of3:1, 1:1,and1:3by volume andfinally in100%xylene,for10min periods at room temperature.Subsequently,the granules were embedded in paraffin(m.p.56e58 C)by serial immer-sion in1:1xylene-paraffin for30min at60 C,followed by 100%paraffin.After solidification in paraffin,8-m m-thick sections were prepared and placed on gelatin-coated micro-scopic slides.Paraffin was removed by immersing the slide in xylene and ethanol for30min each,followed by air-drying of the slides.The three oligonucleotide probes were used for hybridiza-tion(Downing and Nerenberg,2008):FITC-labeled Nso190, which targets the majority of AOB;TRITC-labeled NIT3,which targets Nitrobacter sp.;TRITC-labeled NSR1156,which targets Nitrospira sp.All probe sequences,their hybridization condi-tions,and washing conditions are given in Table1.Oligonu-cleotides were synthesized andfluorescently labeled with fluorochomes by Takara,Inc.(Dalian,China).Hybridizations were performed at46 C for2h with a hybridization buffer(0.9M NaCl,formamide at the percentage shown in Table1,20mM Tris/HCl,pH8.0,0.01% SDS)containing each labeled probe(5ng m LÀ1).After hybrid-ization,unbound oligonucleotides were removed by a strin-gent washing step at48 C for15min in washing buffer containing the same components as the hybridization buffer except for the probes.For detection of all DNA,4,6-diamidino-2-phenylindole (DAPI)was diluted with methanol to afinal concentration of1ng m LÀ1.Cover the slides with DAPI e methanol and incubate for15min at37 C.The slides were subsequently washed once with methanol,rinsed briefly with ddH2O and immediately air-dried.Vectashield(Vector Laboratories)was used to prevent photo bleaching.The hybridization images were captured using a confocal laser scanning microscope (CLSM,Zeiss710).A total of10images were captured for each probe at each class of size.The representative images were selected andfinal image evaluation was done in Adobe PhotoShop.w a t e r r e s e a r c h x x x(2011)1e1033.Results3.1.SBR performance and granule characteristicsDuring the startup period,the reactor removed TOC and NH 4þ-N efficiently.98%of NH 4þ-N and 100%of TOC were removed from the influent by day 3and day 5respectively (Figs.S2,S3,Supporting information ).Removal of TN and TP were lower during this period (Figs.S3,S4,Supporting information ),though the removal of TP gradually improved to 100%removal by day 33(Fig.S4,Supporting information ).To determine the sludge volume index of granular sludge,a settling time of 10min was chosen instead of 30min,because granular sludge has a similar SVI after 60min and after 5min of settling (Schwarzenbeck et al.,2004b ).The SVI 10of the inoculating sludge was 108.2mL g À1.The changing patterns of MLSS and SVI 10in the continuous operation of the SBR are illustrated in Fig.1.The sludge settleability increased markedly during the set-up period.Fig.2reflects the slow andgradual process of sludge granulation,i.e.,from flocculentsludge to granules.3.2.DGGE analysis:AOB communities structure changes during sludge granulationThe results of nested PCR were shown in Fig.S1.The well-resolved DGGE bands were obtained at the representative points throughout the GSBR operation and the patterns revealed that the structure of the AOB communities was dynamic during sludge granulation and stabilization (Fig.3).The community structure at the end of experiment was different from that of the initial pattern of the seed sludge.The AOB communities on day 1showed 40%similarity only to that at the end of the GSBR operation (Table S1,Supporting information ),indicating the considerable difference of AOB communities structures between inoculated sludge and granular sludge.Biodiversity based on the DGGE patterns was analyzed by calculating the Shannon diversity index H as204060801001201401254159738494104115125135147160172188Time (d)S V I 10 (m L .g -1)10002000300040005000600070008000900010000M L S S (m g .L -1)Fig.1e Change in biomass content and SVI 10during whole operation.SVI,sludge volume index;MLSS,mixed liquid suspendedsolids.Fig.2e Variation in granule size distribution in the sludge during operation.d,particle diameter;TSS,total suspended solids.w a t e r r e s e a r c h x x x (2011)1e 104shown in Fig.S5.In the phase of sludge inoculation (before day 38),H decreased remarkably (from 0.94to 0.75)due to the absence of some species in the reactor.Though several dominant species (bands2,7,10,11)in the inoculating sludge were preserved,many bands disappeared or weakened (bands 3,4,6,8,13,14,15).After day 45,the diversity index tended to be stable and showed small fluctuation (from 0.72to 0.82).Banding pattern similarity was analyzed by applying UPGMA (Fig.4)algorithms.The UPGMA analysis showed three groups with intragroup similarity at approximately 67%e 78%and intergroup similarity at 44e 62%.Generally,the clustering followed the time course;and the algorithms showed a closer clustering of groups II and III.In the analysis,group I was associated with sludge inoculation and washout,group IIwithFig.3e DGGE profile of the AOB communities in the SBR during the sludge granulation process (lane labels along the top show the sampling time (days)from startup of the bioreactor).The major bands were labeled with the numbers (bands 1e15).Fig.4e UPGMA analysis dendrograms of AOB community DGGE banding patterns,showing schematics of banding patterns.Roman numerals indicate major clusters.w a t e r r e s e a r c h x x x (2011)1e 105startup sludge granulation and decreasing SVI 10,and group III with a stable system and excellent biomass settleability.In Fig.3,the locations of the predominant bands were excised from the gel.DNA in these bands were reamplified,cloned and sequenced.The comparative analysis of these partial 16S rRNA sequences (Table 2and Fig.S6)revealed the phylogenetic affiliation of 13sequences retrieved.The majority of the bacteria in seed sludge grouped with members of Nitrosomonas and Nitrosospira .Along with sludge granula-tion,most of Nitrosomonas (Bands 2,5,7,9,10,11)were remained or eventually became dominant in GSBR;however,all of Nitrosospira (Bands 6,13,15)were gradually eliminated from the reactor.3.3.Distribution of AOB and NOB in different sized granulesFISH was performed on the granule sections mainly to deter-mine the location of AOB and NOB within the different size classes of granules,and the images were not further analyzed for quantification of cell counts.As shown in Fig.6,in small granules (0.2mm <d <0.45mm),AOB located mainly in the outer part of granular space,whereas NOB were detected only in the core of granules.In medium granules (0.45mm <d <0.6mm),AOB distributed evenly throughout the whole granular space,whereas NOB still existed in the inner part.In the larger granules (d >0.9mm),AOB and NOB were mostly located in the surface area of the granules,and moreover,NOB became rare.4.Discussion4.1.Relationship between granule formation and reactor performanceAfter day 32,the SVI 10stabilized at 20e 35mL g À1,which is very low compared to the values measured for activated sludge (100e 150mL g À1).However,the size distribution of the granules measured on day 32(Fig.2)indicated that only 22%of the biomass was made of granular sludge with diameter largerthan 0.2mm.These results suggest that sludge settleability increased prior to granule formation and was not affected by different particle sizes in the sludge during the GSBR operation.It was observed,however,that the diameter of the granules fluctuated over longer durations.The large granules tended to destabilize due to endogenous respiration,and broke into smaller granules that could seed the formation of large granules again.Pochana and Keller reported that physically broken sludge flocs contribute to lower denitrification rates,due to their reduced anoxic zone (Pochana and Keller,1999).Therefore,TN removal efficiency raises fluctuantly throughout the experiment.Some previous research had demonstrated that bigger,more dense granules favored the enrichment of PAO (Winkler et al.,2011a ).Hence,after day 77,removal efficiency of TP was higher and relatively stable because the granules mass fraction was over 90%and more larger granules formed.4.2.Relationship between AOB communities dynamic and sludge granulationFor granule formation,a short settling time was set,and only particles with a settling velocity higher than 4.5m h À1were retained in the reactor.Moreover,as shown in Fig.1,the variation in SVI 10was greater before day 41(from 108.2mL g À1e 34.1mL g À1).During this phase,large amounts of biomass could not survive in the reactor.A clear shift in pop-ulations was evident,with 58%similarity between days 8and 18(Table S1).In the SBR system fed with acetate-based synthetic wastewater,heterotrophic bacteria can produce much larger amounts of extracellular polysaccharides than autotrophic bacteria (Tsuneda et al.,2003).Some researchers found that microorganisms in high shear environments adhered by extracellular polymeric substances (EPS)to resist the damage of suspended cells by environmental forces (Trinet et al.,1991).Additionally,it had been proved that the dominant heterotrophic species in the inoculating sludge were preserved throughout the process in our previous research (Zhang et al.,2011).It is well known that AOB are chemoau-totrophic and slow-growing;accordingly,numerous AOBw a t e r r e s e a r c h x x x (2011)1e 106populations that cannot become big and dense enough to settle fast were washed out from the system.As a result,the variation in AOB was remarkable in the period of sludge inoculation,and the diversity index of population decreased rapidly.After day 45,AOB communities’structure became stable due to the improvement of sludge settleability and the retention of more biomass.These results suggest that the short settling time (selection pressure)apparently stressed the biomass,leading to a violent dynamic of AOB communities.Further,these results suggest that certain populations may have been responsible for the operational success of the GSBR and were able to persist despite the large fluctuations in pop-ulation similarity.This bacterial population instability,coupled with a generally acceptable bioreactor performance,is congruent with the results obtained from a membrane biore-actor (MBR)for graywater treatment (Stamper et al.,2003).Nitrosomonas e like and Nitrosospira e like populations are the dominant AOB populations in wastewater treatment systems (Kowalchuk and Stephen,2001).A few previous studies revealed that the predominant populations in AOB communities are different in various wastewater treatment processes (Tawan et al.,2005;Thomas et al.,2010).Some researchers found that the community was dominated by AOB from the genus Nitrosospira in MBRs (Zhang et al.,2010),whereas Nitrosomonas sp.is the predominant population in biofilter sludge (Yin and Xu,2009).In the currentstudy,Fig.5e DGGE profile of the AOB communities in different size of granules (lane labels along the top show the range of particle diameter (d,mm)).Values along the bottom indicate the Shannon diversity index (H ).Bands labeled with the numbers were consistent with the bands in Fig.3.w a t e r r e s e a r c h x x x (2011)1e 107sequence analysis revealed that selection pressure evidently effect on the survival of Nitrosospira in granular sludge.Almost all of Nitrosospira were washed out initially and had no chance to evolve with the environmental changes.However,some members of Nitrosomonas sp.have been shown to produce more amounts of EPS than Nitrosospira ,especially under limited ammonia conditions (Stehr et al.,1995);and this feature has also been observed for other members of the same lineage.Accordingly,these EPS are helpful to communicate cells with each other and granulate sludge (Adav et al.,2008).Therefore,most of Nitrosomonas could adapt to this challenge (to become big and dense enough to settle fast)and were retained in the reactor.At the end of reactor operation (day 180),granules with different particle size were sieved.The effects of variation in granules size on the composition of the AOBcommunitiesFig.6e Micrographs of FISH performed on three size classes of granule sections.DAPI stain micrographs (A,D,G);AOB appear as green fluorescence (B,E,H),and NOB appear as red fluorescence (C,F,I).Bar [100m m in (A)e (C)and (G)e (I).d,particle diameter.(For interpretation of the references to colour in this figure legend,the reader is referred to the web version of this article.)w a t e r r e s e a r c h x x x (2011)1e 108were investigated.As shown in Fig.5,AOB communities structures in different size of granules were varied.Although several predominant bands(bands2,5,11)were present in all samples,only bands3and6appeared in the granules with diameters larger than0.6mm.Additionally,bands7and10 were intense in the granules larger than0.45mm.According to Table2,it can be clearly indicated that Nitrosospira could be retained merely in the granules larger than0.6mm.Therefore, Nitrosospira was not present at a high level in Fig.3due to the lower proportion of larger granules(d>0.6mm)in TSS along with reactor operation.DGGE analysis also revealed that larger granules had a greater microbial diversity than smaller ones. This result also demonstrates that more organisms can survive in larger granules as a result of more space,which can provide the suitable environment for the growth of microbes(Fig.6).4.3.Effect of variance in particle size on the distribution of AOB and NOB in granulesAlthough an influence of granule size has been observed in experiments and simulations for simultaneous N-and P-removal(de Kreuk et al.,2007),the effect of granule size on the distribution of different biomass species need be revealed further with the assistance of visible experimental results, especially in the same granular sludge reactors.Related studies on the diversity of bacterial communities in granular sludge often focus on the distribution of important functional bacteria populations in single-size granules(Matsumoto et al., 2010).In the present study,different size granules were sieved,and the distribution patterns of AOB and NOB were explored.In the nitrification processes considered,AOB and NOB compete for space and oxygen in the granules(Volcke et al.,2010).Since ammonium oxidizers have a higheroxygen affinity(K AOBO2<K NOBO2)and accumulate more rapidly inthe reactor than nitrite oxidizers(Volcke et al.,2010),NOB are located just below the layer of AOB,where still some oxygen is present and allows ready access to the nitrite produced.In smaller granules,the location boundaries of the both biomass species were distinct due to the limited existence space provided by granules for both microorganism’s growth.AOB exist outside of the granules where oxygen and ammonia are present.Medium granules can provide broader space for microbe multiplying;accordingly,AOB spread out in the whole granules.This result also confirms that oxygen could penetrate deep into the granule’s core without restriction when particle diameter is less than0.6mm.Some mathematic model also supposed that NOBs are favored to grow in smaller granules because of the higher fractional aerobic volume (Volcke et al.,2010).As shown in the results of the batch experiments(Zhang et al.,2011),nitrite accumulation temporarily occurred,accompanied by the more large gran-ules(d>0.9mm)forming.This phenomenon can be attrib-uted to the increased ammonium surface load associated with larger granules and smaller aerobic volume fraction,resulting in outcompetes of NOB.It also suggests that the core areas of large granules(d>0.9mm)could provide anoxic environment for the growth of anaerobic denitrificans(such as Tb.deni-trificans or Tb.thioparus in Fig.S7,Supporting information).As shown in Fig.2and Fig.S3,the removal efficiency of total nitrogen increased with formation of larger granules.5.ConclusionsThe variation in AOB communities’structure was remarkable during sludge inoculation,and the diversity index of pop-ulation decreased rapidly.Most of Nitrosomonas in the inocu-lating sludge were retained because of their capability to rapidly adapt to the settling e washing out action.DGGE anal-ysis also revealed that larger granules had greater AOB diversity than that of smaller ones.Oxygen penetration was not restricted in the granules of less than0.6mm particle diameter.However,the larger granules(d>0.9mm)can result in the smaller aerobic volume fraction and inhibition of NOB growth.Henceforth,further studies on controlling and opti-mizing distribution of granule size could be beneficial to the nitrogen removal and expansive application of granular sludge technology.AcknowledgmentsThis work was supported by grants from the National Natural Science Foundation of China(No.51108456,50908227)and the National High Technology Research and Development Program of China(No.2009AA06Z312).Appendix.Supplementary dataSupplementary data associated with this article can be found in online version at doi:10.1016/j.watres.2011.09.026.r e f e r e n c e sAdav,S.S.,Lee, D.J.,Show,K.Y.,2008.Aerobic granular sludge:recent advances.Biotechnology Advances26,411e423.APHA-AWWA-WEF,2005.Standard Methods for the Examination of Water and Wastewater,first ed.American Public Health Association/American Water Works Association/WaterEnvironment Federation,Washington,DC.de Bruin,L.M.,de Kreuk,M.,van der Roest,H.F.,Uijterlinde,C., van Loosdrecht,M.C.M.,2004.Aerobic granular sludgetechnology:an alternative to activated sludge?Water Science and Technology49,1e7.de Kreuk,M.,Heijnen,J.J.,van Loosdrecht,M.C.M.,2005.Simultaneous COD,nitrogen,and phosphate removal byaerobic granular sludge.Biotechnology and Bioengineering90, 761e769.de Kreuk,M.,Picioreanu,C.,Hosseini,M.,Xavier,J.B.,van Loosdrecht,M.C.M.,2007.Kinetic model of a granular sludge SBR:influences on nutrient removal.Biotechnology andBioengineering97,801e815.Downing,L.S.,Nerenberg,R.,2008.Total nitrogen removal ina hybrid,membrane-aerated activated sludge process.WaterResearch42,3697e3708.Erguder,T.H.,Boon,N.,Vlaeminck,S.E.,Verstraete,W.,2008.Partial nitrification achieved by pulse sulfide doses ina sequential batch reactor.Environmental Science andTechnology42,8715e8720.w a t e r r e s e a r c h x x x(2011)1e109。
- 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)。
kernel set
indicators
classes
050010001500
0
10
20
30
40
50
60
Iteration
L
050010001500
0
5
10
15
20
Iteration
K
−2−1012
−3
−2
−1
0
1
2
3
x
1
x
2
−2−1012
−3
−2
−1
0
1
2
3
x
1
x
2
050010001500
0
100
200
300
400
500
600
700
Iteration
L
050010001500
0
5
10
15
20
K
Iteration
02004006008001000
1
2
t (seconds)
C
l
a
s
s
l
a
b
e
l
050010001500
0
2000
4000
6000
8000
10000
12000
14000
Iteration
L
050010001500
0
5
10
15
20
K
Iteration
Class label