LED-FDTD LED时域有限差分方法
时域有限差分法(FDTD算法)的基本原理及仿真
时域有限差分法(FDTD算法)的基本原理及仿真时域有限差分法(FDTD 算法)时域有限差分法是1966年K.S.Yee 发表在AP 上的一篇论文建立起来的,后被称为Yee 网格空间离散方式。
这种方法通过将Maxwell 旋度方程转化为有限差分式而直接在时域求解, 通过建立时间离散的递进序列, 在相互交织的网格空间中交替计算电场和磁场。
FDTD 算法的基本思想是把带时间变量的Maxwell 旋度方程转化为差分形式,模拟出电子脉冲和理想导体作用的时域响应。
需要考虑的三点是差分格式、解的稳定性、吸收边界条件。
有限差分通常采用的步骤是:采用一定的网格划分方式离散化场域;对场内的偏微分方程及各种边界条件进行差分离散化处理,建立差分格式,得到差分方程组;结合选定的代数方程组的解法,编制程序,求边值问题的数值解。
1.FDTD 的基本原理FDTD 方法由Maxwell 旋度方程的微分形式出发,利用二阶精度的中心差分近似,直接将微分运算转换为差分运算,这样达到了在一定体积内和一段时间上对连续电磁场数据的抽样压缩。
Maxwell 方程的旋度方程组为:E E H σε+∂∂=⨯∇t H HE m tσμ-∂∂-=⨯∇ (1) 在直角坐标系中,(1)式可化为如下六个标量方程:⎪⎪⎪⎪⎭⎪⎪⎪⎪⎬⎫+∂∂=∂∂-∂∂+∂∂=∂∂-∂∂+∂∂=∂∂-∂∂z z x y y y z x x x yz E t E y H x H E t E x H z H E t E z H y H σεσεσε,⎪⎪⎪⎪⎭⎪⎪⎪⎪⎬⎫-∂∂-=∂∂-∂∂-∂∂-=∂∂-∂∂-∂∂-=∂∂-∂∂z m zx y y m y z x x m x y z H t H y E x E H t H x E z E H t H z E y E σμσμσμ (2)上面的六个偏微分方程是FDTD 算法的基础。
Yee 首先在空间上建立矩形差分网格,在时刻t n ∆时刻,F(x,y,z)可以写成),,(),,,(),,,(k j i F t n z k y j x i F t z y x F n =∆∆∆∆= (3)用中心差分取二阶精度: 对空间离散:()[]2),,21(),,21(),,,(x O xk j i F k j i F x t z y x F n n xi x ∆+∆--+≈∂∂∆= ()[]2),21,(),21,(),,,(y O yk j i F k j i F y t z y x F n n yj y ∆+∆--+≈∂∂∆= ()[]2)21,,()21,,(),,,(z O zk j i F k j i F z t z y x F n n zk z ∆+∆--+≈∂∂∆=对时间离散:()[]22121),,(),,(),,,(t O tk j i F k j i F t t z y x F n n tn t ∆+∆-≈∂∂-+∆= (4) Yee 把空间任一网格上的E 和H 的六个分量,如下图放置:oyxzEyHzExEzHxEyEyEzEx HyEzEx图1 Yee 氏网格及其电磁场分量分布在FDTD 中,空间上连续分布的电磁场物理量离散的空间排布如图所示。
时域有限差分法
时域有限差分法时域有限差分法(TimeDomainFiniteDifferenceMethod,简称TD-FDM)是数值分析领域中非常重要的一种数值计算方法,它是利用有限差分法对时域偏微分方程(PDE)进行求解的一种方法,其应用范围十分广泛,是在工程和科学领域中应用最多的计算方法之一。
时域有限差分法可以精确表示任意时域偏微分方程的解,但是由于求解过程中存在计算量大、精度低、收敛慢等问题,其计算效率和精度也有限。
因此,人们必须采取有效的方法来提高此类方法的精度和计算效率,增强其在工程和科学领域的应用价值。
时域有限差分法的原理很简单,即将偏微分方程的解以一系列有规律的离散点表示,再利用有限差分对偏微分方程进行求解。
它主要包括三个部分:数值模型构建、数值计算和数值结果分析。
首先,根据时域偏微分方程的类型及物理本质,构建与之对应的数值模型,采用有限差分形式表达偏微分方程,并根据时域偏微分方程的解特性对有限差分方程进行增强。
然后,构建时域有限差分的计算框架,利用计算机编程语言(如C++、Fortran、Python等)实现数值计算,采用常用的多项式插值和求解算法(如牛顿迭代法、拟牛顿法等)实现精确计算。
最后,利用计算机绘图软件对所得到的数值结果进行分析,以评估结果的准确性,并做出相应的修改和优化。
时域有限差分法的应用非常广泛,它可以用于各种工程领域,如稳态和不稳态流动场的求解,声学学中的各类传播现象的模拟,热传导的分析等。
此外,时域有限差分法在一些科学领域也有很大的应用,如量子力学中电子能级结构、原子结构的计算,核物理中文中阳离子反应剂度模拟,生物学中细胞动力学模型仿真等等。
近年来,随着计算机技术的进一步发展,出现了许多新的发展方向:从传统的有限差分法到基于保守型的计算方法,从基于有穷元的数值模拟方法到超差分法,从动态网格特定的方法到基于机器学习的计算方法。
所有这些方法都可以用于处理更复杂的时域偏微分方程,提高精度和计算效率。
时域有限差分法二维
时域有限差分法二维1. 引言时域有限差分法(Finite Difference Time Domain, FDTD)是一种常用的数值计算方法,用于求解电磁场在时域中的传播和辐射问题。
本文将以二维情况为例,深入探讨时域有限差分法的原理和应用。
通过本文的介绍和解读,您将更全面地理解这一方法,并能够灵活应用于相关领域。
2. 时域有限差分法简介2.1 原理概述时域有限差分法是一种迭代求解偏微分方程的方法,通过将时域和空间离散化,将连续问题转化为离散问题。
在二维情况下,假设空间网格分辨率为Δx和Δy,时间步长为Δt。
根据电磁场的麦克斯韦方程组,可以利用中心差分公式进行离散化计算,得到求解方程组的更新方程。
2.2 空间离散化对于二维情况,空间离散化可以采用正交网格或非正交网格。
常见的正交网格包括方形格点、Yee网格等,而非正交网格则具有更灵活的形态。
根据需要和应用场景,选择合适的离散化方法对问题进行求解。
2.3 时间离散化时间离散化主要有显式和隐式两种方法。
显式方法将时间推进方程展开成前一时刻的电场和磁场与当前时刻的源项之间的关系,容易计算但对时间步长有限制;隐式方法则是通过迭代或矩阵计算求解当前时刻的电场和磁场。
3. 时域有限差分法的应用领域时域有限差分法广泛应用于电磁场传播和辐射问题的数值模拟中。
以下是几个典型的应用领域:3.1 辐射问题时域有限差分法可以模拟电磁波在空间中的辐射传播过程。
可以用于分析天线的辐射特性,设计无线通信系统的天线,或者分析电磁波在无线电频段的传播情况。
3.2 波导问题对于波导结构,时域有限差分法可以求解其模式、传输特性等问题。
波导结构广泛应用于光子学器件、微波器件等领域,时域有限差分法为建立数值模型和解析波导特性提供了一种有效的数值计算手段。
3.3 散射问题时域有限差分法在散射问题的数值模拟中也有重要应用。
通过模拟散射体与电磁波的相互作用过程,可以研究和分析散射体的散射特性,例如雷达散射截面的计算、微波散射问题等。
时域有限差分法介绍
时域有限差分法介绍
时域有限差分法(Finite Difference Time Domain, FDTD)是
一种数值求解电磁波在时域中传播的方法。
它通过将空间和时间连续
性方程离散化,将偏微分方程转化为差分方程,并使用差分法来近似
求解波动方程。
时域有限差分法可以用于研究不同频率和波长的电磁波在各向同性、各向异性以及具有非线性、色散等特性的介质中的传播和相互作用。
它广泛应用于光学和电磁学领域中,可用于模拟光纤、微波器件、天线、光子晶体、超材料等的性能。
该方法的基本思想是将空间划分为离散的单元,称为网格,其中
包含了电场、磁场、电流和电荷等物理量。
通过对空间坐标和时间进
行离散化,可以将连续的偏微分方程转化为差分方程。
具体地,通过
泰勒展开将时域和空域的导数转化为有限差分的形式。
在时域有限差分法中,电场和磁场被分别定义在正方形的网格节
点上。
通过应用麦克斯韦方程组的差分形式,可以得到给定时间步长
的下一个时间步的电场和磁场值。
这些值可以根据初始条件和边界条
件进行更新。
时域有限差分法具有较好的稳定性和精度,可以模拟各种复杂的
电磁现象。
然而,它在处理边界条件和非均匀介质等问题时存在一些
困难。
因此,研究者们提出了各种改进的时域有限差分法,以提高其
适用性和效率。
时域有限差分法
时域有限差分法
时域有限差分法的基本思想是用中心差商代替场量对时间和空间的一阶偏微商, 通过在时域的递推模拟波的传播过程, 从而得出场分布。
它最早由K.S.Yee 于1966 年提出,在此之后的20 年内,其研究进展缓慢,只是在电磁散射、电磁兼容领域有一些初步的应用。
自80 年代末,时域有限差分法成为电磁场数值计算的重要方法之一。
在声学数值计算中,时域有限差分法已应用于水声学、噪声控制及室内声学等方面的数值模拟。
时域有限差分法(Finite difference time domainmethod,FDTD)直接离散时域波动方程,不需要任何形式的导出方程,故不会因为数学模型而限制其应用范围。
它的差分格式中包含有介质的参量,只须赋予各网格相应的参量,就能模拟各种复杂的结构,这是时域有限差分法的一个突出优点。
另外,由于时域有限差分法采用步进法进行计算,故能很容易地实现各种复杂时域宽带信号的模拟,而且可以非常方便地获得空间某一点的时域信号波形。
电磁波时域有限差分方法
电磁波时域有限差分方法
电磁波时域有限差分法(Finite-Difference Time-Domain Method, FDTD)是一种求解电磁学问题的常用数值方法。
它由Yee在1966年首次提出,可用于求解复杂三维电磁场交互作用的问题,如,电磁波、磁致传导、微波加热、能量传输、电磁辐射等。
相比其它数值方法,FDTD方法求解算例更为精确,具有以下特点:
1. TDTD方法是在时域上,而非在频域中,因此可以方便地处理暂态和复杂变化的电磁场。
2. FDTD方法可以通过改变差分格式和计算网格或计算量来获得更加精确的结果。
3. FDTD方法可以数值模拟出任何电磁场的行为,并且可以得到高质量的结果,而且不受物理规律的限制。
4. 可以自动识别模型中的隐藏材料特性,并增强模型的实用性。
5. FDTD方法可以结合有限体积法(FVM)和有限元法(FEM),提高模型的精度,并减少工作量。
6. 较少的内存要求,使FDTD方法更适用于工程应用。
FDTD方法在处理复杂电磁场时,有时会导致计算窗口大小,以及时间分辨率的降低,因此,要想获得较为准确的结果,就要采取足够的计算网格,以及足够高的时间分辨率。
FDTD时域有限差分法
对时间离散:
(2)
FDTD基本原理(续)
9
为了满足(1)式空间精度的要求,并满足(2)式,Yee 把空间任一网格上的E和H的六个分量,如下图放置:
Yee把E 和H 在时间长相差半个步长计算(为了满足精度的要求)。
FDTD基本原理(续)
10
根据这一原则可以写出六个差分方程:
每个网格点上的各场分量的新值依赖于该点在前一时间步长时刻 的值,即该点周围的邻近点上另一场量在早半个时间步长时的值。 因此任一时刻可一次算出一个点,并行算法可计算出多个点。通 过这些运算可以交替算出电场磁场在各个时间步的值。
C:为光速,自由空间中: c
数值色散
14
• 产生原因
–FDTD网格中,会导致数字波模在网格中发生改变,这种改 变是由于计算网格本身引起的,而非物理因素,所以必须 考虑
• 适当选取时间步长,空间步长,传播方向,可以得到 理想情况
–3-D方形网格:取波沿对角线传播 (数值稳定的极限状态),可得理想色散关系。 –2-D方形网格:也是沿对角线传播, (也是数值稳定的极限状态) –1-D网格 (数值稳定的极限状态)
参考文献
21
• 电磁波时域有限差分方法(第二版),葛德彪, 闫玉波,西安电子科技大学出版社 • 工程电磁场数值计算,倪光正
22
练习要求:
fdtd 有限差分法加源 面电流源和线电流源
FDTD (有限差分时域法)是一种对计算电动力学建模的数值分析技术,由K.S.Yee在1966年提出。
它本质上是将随时间变化的Maxwell 旋度方程更改为离散差分形式,将连续的空间分成有限的网格进行计算。
在FDTD方法中,可以通过在特定位置设置源来模拟电流。
这些源可以包括面电流源和线电流源。
面电流源可以模拟一个二维平面上分布的电流,而线电流源则可以模拟- -维物体上的电流分布。
对于面电流源,可以在特定平面上施加电流,电流的分布可以在该平面上进行定义。
例如,可以在平面上施加一个恒定的电流密度,或者定义一个更复杂的电流分布模式。
对于线电流源,可以在特定线上施加电流,电流的分布可以在该线上进行定义。
例如,可以在线上施加一个恒定的电流强度,或者定义-个更复杂的电流分布模式。
在FDTD方法中,通过在特定的时间步长内对源进行更新,并使用差分方程来更新电场和磁场,可以模拟电磁波的传播和散射等物理现象。
这种方法的优点是它可以模拟复杂的电磁场分布,并且可以很容易地处理各种材料和边界条件。
总的来说,FDTD是一种强大的数值分析工具,可以用来模拟和分析电磁波在复杂环境中的传播和散射等物理现象。
通过合理设置面电流源和线电流源等源项,可以模拟和分析各种不同情况下的电磁场分布和传播特性。
时域有限差分方法
时域有限差分方法
时域有限差分方法(FDTD)是一种数值求解电磁场问题的方法,适用于计算复杂的电磁现象。
该方法将电磁场方程离散化为差分形式,然后通过不断迭代求解差分方程,得到电磁场在时域上的时变分布。
具体来说,FDTD方法将空间和时间分割成网格,然后在每个网格点上估计电磁场的值。
通过使用差分方程,可以将电场和磁场的时变分布递推到下一个时间步。
一般而言,FDTD方法采用中心差分形式的差分方程,以提高数值解的稳定性和精度。
FDTD方法的主要优点是适用于计算非线性、吸收、散射等复杂电磁现象。
由于差分形式的方程可以直接计算,相比其他数值方法(如有限元方法和边界元方法),FDTD方法具有较高的计算速度。
然而,FDTD方法也存在一些限制。
由于需要将空间和时间分割为网格,因此对于复杂几何形状和大尺寸问题,需要较大的计算资源和内存。
此外,FDTD方法对吸收边界条件的处理也比较复杂,需要采用合适的数值技巧来避免误差累积。
总的来说,FDTD方法是一种广泛应用于电磁场问题求解的数值方法,具有较高的计算速度和适用性。
在实际应用中,可以结合其他方法或技术对其进行改进和优化,以适应各种特定问题的求解需求。
时域有限差分方法
时域有限差分方法
《时域有限差分方法》
嘿,你知道吗,有一种超厉害的方法叫时域有限差分方法!这可真是个神奇的玩意儿。
想象一下,我们要研究那些看不见摸不着的电磁波啊之类的东西。
以前可麻烦了,但有了时域有限差分方法,就好像打开了一扇新的大门。
它是怎么工作的呢?简单来说,就是把我们要研究的区域划分成很多很多小格子,就像一个大拼图一样。
然后呢,通过计算这些小格子之间的变化,来了解整个区域的情况。
这个方法的好处可多啦!它能处理各种复杂的情况,不管是奇形怪状的物体,还是变化多端的环境。
而且,它很直观,让我们能清楚地看到电磁波是怎么传播、怎么变化的。
在实际应用中,时域有限差分方法可太有用了。
比如在通信领域,它能帮助我们设计更好的天线,让信号传输得更远更稳定。
在雷达系统中,它能让我们更准确地探测目标。
我觉得时域有限差分方法真的是一项非常了不起的技术,给我们探索和理解各种物理现象带来了巨大的帮助。
时域有限差分法(FDTD算法)的基本原理及仿真
时域有限差分法(FDTD算法)的基本原理及仿真时域有限差分法(FDTD 算法)时域有限差分法是1966年K.S.Yee 发表在AP 上的一篇论文建立起来的,后被称为Yee 网格空间离散方式。
这种方法通过将Maxwell 旋度方程转化为有限差分式而直接在时域求解, 通过建立时间离散的递进序列, 在相互交织的网格空间中交替计算电场和磁场。
FDTD 算法的基本思想是把带时间变量的Maxwell 旋度方程转化为差分形式,模拟出电子脉冲和理想导体作用的时域响应。
需要考虑的三点是差分格式、解的稳定性、吸收边界条件。
有限差分通常采用的步骤是:采用一定的网格划分方式离散化场域;对场内的偏微分方程及各种边界条件进行差分离散化处理,建立差分格式,得到差分方程组;结合选定的代数方程组的解法,编制程序,求边值问题的数值解。
1.FDTD 的基本原理FDTD 方法由Maxwell 旋度方程的微分形式出发,利用二阶精度的中心差分近似,直接将微分运算转换为差分运算,这样达到了在一定体积内和一段时间上对连续电磁场数据的抽样压缩。
Maxwell 方程的旋度方程组为:E E H σε+∂∂=⨯∇t H HE m tσμ-∂∂-=⨯∇ (1) 在直角坐标系中,(1)式可化为如下六个标量方程:⎪⎪⎪⎪⎭⎪⎪⎪⎪⎬⎫+∂∂=∂∂-∂∂+∂∂=∂∂-∂∂+∂∂=∂∂-∂∂z z x y y y z x x x yz E t E y H x H E t E x H z H E t E z H y H σεσεσε,⎪⎪⎪⎪⎭⎪⎪⎪⎪⎬⎫-∂∂-=∂∂-∂∂-∂∂-=∂∂-∂∂-∂∂-=∂∂-∂∂z m zx y y m y z x x m x y z H t H y E x E H t H x E z E H t H z E y E σμσμσμ (2)上面的六个偏微分方程是FDTD 算法的基础。
Yee 首先在空间上建立矩形差分网格,在时刻t n ∆时刻,F(x,y,z)可以写成),,(),,,(),,,(k j i F t n z k y j x i F t z y x F n =∆∆∆∆= (3)用中心差分取二阶精度: 对空间离散:()[]2),,21(),,21(),,,(x O xk j i F k j i F x t z y x F n n xi x ∆+∆--+≈∂∂∆= ()[]2),21,(),21,(),,,(y O yk j i F k j i F y t z y x F n n yj y ∆+∆--+≈∂∂∆= ()[]2)21,,()21,,(),,,(z O zk j i F k j i F z t z y x F n n zk z ∆+∆--+≈∂∂∆=对时间离散:()[]22121),,(),,(),,,(t O tk j i F k j i F t t z y x F n n tn t ∆+∆-≈∂∂-+∆= (4) Yee 把空间任一网格上的E 和H 的六个分量,如下图放置:oyxzEyHzExEzHxEyEyEzEx HyEzEx图1 Yee 氏网格及其电磁场分量分布在FDTD 中,空间上连续分布的电磁场物理量离散的空间排布如图所示。
时域有限差分法FD-TD
内容提要
• FD-TD法基本要素
– 物理意义 – Yee单元 – FD-TD法的基本算式 – 数值稳定性及数值色散 – Maxwell方程的差分方程 – 网格的大小 – 吸收边界条件 – 入射波源的设置 – 远场计算
• FD-TD法的特点 • 算例
航空宇航学院
• 物理意义
– 直接描述了电磁波的传播过程,从时间t=0由波 源开始,通过执行每一空间网格的有限差分计算使 波所在模拟空间传播开来,并与在空间中模拟的结 构相互作用。
航空宇航学院
• Yee单元
航空宇航学院
FD-TD法的基本算式
笛卡尔坐标中Maxwell方程
∂H x 1 ∂E x ∂E z = ( − − ρ ' H x) ∂t µ ∂z ∂y
∂H y ∂t = 1 ∂E z ∂E x ( − − ρ ' H y) µ ∂x ∂z
∂E x 1 ∂H z ∂H y − − σE x ) = ( ε ∂y ∂z ∂t
F n (i, j , k ) = F (i∆x, j∆y , k∆z , n∆t )
∂F n (i, j , k ) F n (i + 1 / 2, j , k ) − F n (i − 1 / 2, j , k ) = + order (∆x 2 ) ∂x ∆x
∂F n (i, j , k ) F n +1/ 2 (i, j , k ) − F n −1/ 2 (i, j , k ) = + order (∆t 2 ) ∂t ∆t
航空宇航学院
时域有限差分法( FD-TD )
Finite Difference-Time Domain Method
航空宇航学院
时域有限差分有限元
时域有限差分有限元
时域有限差分(FDTD)和有限元法(FEM)是两种常用的数值模
拟方法,用于求解时域中的波动现象和电磁场问题。
它们在工程学、物理学和地球科学等领域都有广泛的应用。
首先,让我们从时域有限差分(FDTD)方法开始。
FDTD方法是
一种数值求解Maxwell方程组的离散化方法,它将时域Maxwell方
程组转化为差分形式,通过在空间和时间上进行离散化,将连续的
时域问题转化为离散的网格问题。
FDTD方法的优点包括易于理解和
实现、适用于各种介质和边界条件,能够模拟宽频段的波动现象等。
在电磁场、光学、天线设计等领域得到了广泛的应用。
其次,让我们来看看有限元法(FEM)。
有限元法是一种广泛应
用的数值分析方法,用于求解偏微分方程和变分问题。
在时域中,
有限元法可以用于求解Maxwell方程组、热传导方程等问题。
有限
元法将求解区域分割成有限数量的单元,通过建立单元之间的关系,建立整个系统的离散方程,然后通过数值方法求解得到近似解。
有
限元法的优点包括适用于复杂几何形状、能够处理各向异性材料、
可以考虑不同类型的边界条件等。
综上所述,时域有限差分和有限元法都是重要的数值模拟方法,在不同的领域有着广泛的应用。
它们各自有着特点和适用范围,选
择合适的方法取决于具体的求解问题和模拟需求。
在工程实践中,
通常需要根据具体情况来选择合适的数值模拟方法,以获得准确的
仿真结果。
电磁波时域有限差分方法(FDTD)
������ ∆������ ≤ 12
������为无色散介质中, ������ = 2������/������。对于非单 色波的时域脉冲信号,应以信号款到中 所对应的上限频率的波长������������������������来代替
考虑一维波动方程的一般形式:
������2������ 1 ������2������ ������������2 − ������2 ������������2 = 0
有限差分的二阶导数近似:
将(11)带入(12)得:
������2������ ������ ������ + ∆������ − 2������ ������ + ������(������ − ∆������)
������������2 ≈
(∆������)2
(1-12)
������2������ ������������2
(1-25)
为了减小上式所对应的数值色散,除了选 择空间离散∆������外,对于时间离散有同样 的选择,即:
差分近似关系:
������∆������ ������ 2 ≤ 12 或
������ ∆������ ≤ 12
������������������2(������������2∆������) (∆2������)2
������ ������∆������ ≤
3
(2) 二维取∆������ = ∆������ = ������时:
������ ������∆������ ≤
2
(1-15) (1-16) (1-17) (1-18)
(3) 一维时:
������∆������ ≤ ∆������
(1-19)
时域有限差分法
3 时域有限差分法(FDTD )1966年K.S. Yee 发表了时域有限差分法(Finite Difference -Time Domain ,简记FDTD)的奠基性论文[1],之后在很长一段时间内,这一思想没有引起电磁理论界的足够重视。
直到七十年代末八十年代初,在A. Taflove [2]、K.S. Kunz [3]和R. Holland [4]等学者的推进下,这一方法才逐渐走向成熟并得到广泛的研究和应用。
时域有限差分法的原理非常简单,就是直接将时域Maxwell 方程组的两个旋度方程中关于空间变量和时间变量的偏导数用差商近似,从而转换为离散网格节点上的时域有限差分方程。
加入时域脉冲激励后,在时间上迭代就可直观地模拟出脉冲在求解区域上传播、反射和散射的过程,进而采用FFT 将时域响应变换到频域就可获得所希望的各种电参数,如无源电路的散射参数、天线的辐射方向图和输入阻抗、散射体的雷达散射截面(RCS)等。
随着FDTD 方法的迅猛发展,新的处理方法和技术不断涌现。
其中,子网格模型技术是用子网格或细网格划分薄片、裂缝和导线,其余部分用粗网格进行划分,以便在不显著增加计算时间的基础上提高计算精度;非正交和广义正交曲线网格技术适应于各种结构形状,可以模拟各种复杂的结构;非均匀正交网格技术在复杂结构区域或在场量快变化区域采用细网格,而在其它地方用粗网格,可以兼顾计算时间、存储量和计算精度;回路积分法从积分形式的Faraday 定律和推广的Ampere 定律出发导出回路积分表示的差分格式,使之适用于任意形状的网格结构;外推技术从前面有限时间步的瞬时响应外推以后瞬时响应以大量节省计算时间;网格压缩模型技术用于导波结构分析,通过解析处理,可将传播方向的网格压缩为零。
此外还有,超吸收边界条件技术、色散吸收边界条件技术、完全匹配层吸收边界条件技术、多分辨率技术、伪谱技术、及混合显-隐格式算法等。
新方法与技术的发展迅速扩大了时域有限差分法的应用范围,该方法不仅在目标电磁散射问题,而且在电磁兼容预测、微波电路分析、天线辐射特性计算和生物电磁学研究等方面中都获得了广泛的应用。
时域有限差分法介绍
时域有限差分法(FDTD)是求解电磁波传输问题的一种数值模拟方法。
它是一种在时域内对波动方程进行差分逼近的方法,通过迭代求解离散化后的波动方程,可以得到
电磁波在空间和时间上的分布情况,进而预测电磁波传输的行为。
时域有限差分法主要包括以下几个步骤:
1. 空间离散化:将待求解区域划分为若干个小网格,然后在每个网格内选择一个计算点,利用有限差分法对该点的电场、磁场进行离散化处理,建立电场和磁场的离散计
算模型。
2. 时间推进:时间也进行离散化,将求解时间区间等分成若干个小时间步长,然后依
次求解每个时间步长中(t+Δt)时刻的电场、磁场分布情况。
3. 边界条件处理:根据物理边界条件,对离散化后的电场、磁场进行边界条件处理,
使其在边界处满足边界条件。
4. 迭代求解:在时间和空间上依次迭代求解电场、磁场的分布情况,直到满足设定的
收敛条件或达到一定的迭代次数为止。
时域有限差分法是求解电磁波传输问题的常用方法,它具有以下几个优点:
1. 可以模拟任意形状的物体和复杂的介质结构,适用于不规则和非线性介质。
2. 空间和时间离散化均匀,计算精度高,能够得到电磁波在空间和时间上的分布情况,提供更加详细的仿真结果。
3. 算法简单,易于实现和计算,适用于大规模计算和高性能计算。
4. 可以模拟各种类型的电磁波,如光、微波、射频信号等,广泛应用于光学、无线通信、雷达、医学影像等领域。
总的来说,时域有限差分法是一种有效的求解电磁波传输问题的数值模拟方法,具有
广泛的应用前景。
- 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)。
Efficiency enhancement of homoepitaxialInGaN/GaN light-emitting diodes on free-standing GaN substrate with double embeddedSiO2 photonic crystalsTongbo Wei,* Ziqiang Huo, Yonghui Zhang, Haiyang Zheng, Yu Chen, Jiankun Yang, Qiang Hu, Ruifei Duan, Junxi Wang, Yiping Zeng, and Jinmin Li Semiconductor Lighting Technology Research and Development Center, Institute of Semiconductors, ChineseAcademy of Sciences, Beijing 100083, China*tbwei@Abstract: Homoepitaxially grown InGaN/GaN light emitting diodes(LEDs) with SiO2 nanodisks embedded in n-GaN and p-GaN as photoniccrystal (PhC) structures by nanospherical-lens photolithography arepresented and investigated. The introduction of SiO2 nanodisks doesn’tproduce the new dislocations and doesn’t also result in the electricaldeterioration of PhC LEDs. The light output power of homoepitaxial LEDswith embedded PhC and double PhC at 350 mA current is increased by29.9% and 47.2%, respectively, compared to that without PhC. Thecorresponding light radiation patterns in PhC LEDs on GaN substrate showa narrow beam shape due to strong guided light extraction, with a viewangle reduction of about 30°. The PhC LEDs are also analyzed in detail byfinite-difference time-domain simulation (FDTD) to further reveal theemission characteristics.©2014 Optical Society of AmericaOCIS codes: (230.0230) Optical devices; (230.3670) Light-emitting diodes; (160.5298)Photonic crystals; (220.4241) Nanostructure fabrication.References and links1. B. Monemar and B. E. Sernelius, “Defect related issues in the “current roll-off” in InGaN based light emittingdiodes,” Appl. Phys. Lett. 91(18), 181103 (2007).2. G. Verzellesi, D. Saguatti, M. Meneghini, F. Bertazzi, M. Goano, G. Meneghesso, and E. Zanoni, “Efficiencydroop in InGaN/GaN blue light-emitting diodes: Physical mechanisms and remedies,” J. Appl. Phys. 114(7), 071101 (2013).3. K. Akita, T. Kyono, Y. Yoshizumi, H. Kitabayashi, and K. Katayama, “Improvements of external quantumefficiency of InGaN-based blue light-emitting diodes at high current density using GaN substrates,” J. Appl.Phys. 101(3), 033104 (2007).4. Y. Yang, X. A. Cao, and C. H. Yan, “Rapid efficiency roll-off in high-quality green light-emitting diodes onfreestanding GaN substrates,” Appl. Phys. Lett. 94(4), 041117 (2009).5. C.-L. Chao, R. Xuan, H.-H. Yen, C.-H. Chiu, Y.-H. Fang, Z.-Y. Li, B.-C. Chen, C.-C. Lin, C.-H. Chiu, Y.-D.Guo, J.-F. Chen, and S.-J. Cheng, “Reduction of Efficiency Droop in InGaN Light-Emitting Diode Grown on Self-Separated Freestanding GaN Substrates,” IEEE Photon. Technol. Lett. 23(12), 798–800 (2011).6. M. J. Cich, R. I. Aldaz, A. Chakraborty, A. David, M. J. Grundmann, A. Tyagi, M. Zhang, F. M. Steranka, andM. R. Krames, “Bulk GaN based violet light-emitting diodes with high efficiency at very high current density,”Appl. Phys. Lett. 101(22), 223509 (2012).7. X. A. Cao, S. F. LeBoeuf, M. P. D’Evelyn, S. D. Arthur, J. Kretchmer, C. H. Yan, and Z. H. Yang, “Blue andnear-ultraviolet light-emitting diodes on free-standing GaN substrates,” Appl. Phys. Lett. 84(21), 4313 (2004). 8. Y. J. Zhao, J. Sonoda, C.-C. Pan, S. Brinkley, I. Koslow, K. Fujito, H. Ohta, S. P. DenBaars, and S. Nakamura,“30-mW-class high-power and high-efficiency blue (1011) semipolar InGaN/GaN light-emitting diodes obtained by backside roughening technique,” Appl. Phys. Express 3, 102101 (2010).9. Y.-K. Fu, B.-C. Chen, Y.-H. Fang, R.-H. Jiang, Y.-H. Lu, R. Xuan, K.-F. Huang, C.-F. Lin, Y.-K. Su, J.-F. Chen,and C.-Y. Chang, “Study of InGaN-based light-emitting diodes on a roughened backside GaN substrate by a chemical wet-etching process,” IEEE Photon. Technol. Lett. 23(19), 1373–1375 (2011).#209568 - $15.00 USD Received 4 Apr 2014; revised 23 May 2014; accepted 26 May 2014; published 2 Jun 2014 (C) 2014 OSA30 June 2014 | Vol. 22, No. S4 | DOI:10.1364/OE.22.0A1093 | OPTICS EXPRESS A109310. S. E. Brinkley, C. L. Keraly, J. Sonoda, C. Weisbuch, J. S. Speck, S. Nakamura, and S. P. DenBaars, “ChipShaping for Light Extraction Enhancement of Bulk c-Plane Light-Emitting Diodes,” Appl. Phys. Express 5(3), 032104 (2012).11. B. Sun, L. X. Zhao, T. B. Wei, X. Y. Yi, Z. Q. Liu, G. H. Wang, and J. M. Li, “Shape designing for lightextraction enhancement bulk-GaN light-emitting diodes,” J. Appl. Phys. 113(24), 243104 (2013).12. T. B. Wei, K. Wu, Y. Chen, J. Yu, Q. Yan, Y. Y. Zhang, R. Duan, J. Wang, Y. Zeng, and J. M. Li, “Improvinglight output of vertical-stand-type InGaN light-emitting diodes grown on a free-standing GaN substrate with self-assembled conical arrays,” IEEE Electron Device Lett. 33(6), 857–859 (2012).13. C. Wiesmann, K. Bergenek, N. Linder, and U. T. Schwarz, “Photonic crystal LEDs–designing light extraction,”Laser Photon. Rev. 3(3), 262–286 (2009).14. Y.-J. Kim, M.-K. Kwon, K.-S. Lee, S.-J. Park, S. H. Kim, and K.-D. Lee, “Enhanced light extraction from GaN-based green light-emitting diode with photonic crystal,” Appl. Phys. Lett. 91(18), 181109 (2007).15. H. W. Huang, J. K. Huang, K. Y. Lee, C. F. Lin, and H. C. Guo, “Light-output-power enhancement of GaN-based light-emitting diodes on an n-GaN layer using a SiO2 photonic quasi-crystal overgrowth,” IEEE Electron Device Lett. 31(6), 573–575 (2010).16. K. H. Li and H. W. Choi, “InGaN light-emitting diodes with indium-tin-oxide photonic crystal current-spreadinglayer,” J. Appl. Phys. 110(5), 053104 (2011).17. J. Jewell, D. Simeonov, S.-C. Huang, Y.-L. Hu, S. Nakamura, J. Speck, and C. Weisbuch, “Double embeddedphotonic crystals for extraction of guided light in light-emitting diodes,” Appl. Phys. Lett. 100(17), 171105(2012).18. A. David, B. Moran, K. McGroddy, E. Matioli, E. L. Hu, S. P. DenBaars, S. Nakamura, and C. Weisbuch,“GaN/InGaN light emitting diodes with embedded photonic crystal obtained by lateral epitaxial overgrowth,”Appl. Phys. Lett. 92(11), 113514 (2008).19. T. B. Wei, K. Wu, D. Lan, Q. F. Yan, Y. Chen, C. X. Du, J. X. Wang, Y. P. Zeng, and J. M. Li, “Selectivelygrown photonic crystal structures for high efficiency InGaN emitting diodes using nanospherical-lenslithography,” Appl. Phys. Lett. 101(21), 211111 (2012).20. D. S. Wuu, W. K. Wang, K. S. Wen, S. C. Huang, S. H. Lin, S. Y. Huang, C. F. Lin, and R. H. Horng, “Defectreduction and efficiency improvement of near-ultraviolet emitters via laterally overgrown GaN on aGaN/patterned sapphire template,” Appl. Phys. Lett. 89(16), 161105 (2006).21. Q. Shan, D. S. Meyaard, Q. Dai, J. Cho, E. F. Schubert, J. K. Son, and C. Sone, “Transport-mechanism analysisof the reverse leakage current in GaInN light-emitting diodes,” Appl. Phys. Lett. 99(25), 253506 (2011).22. J. J. Wierer, M. R. Krames, J. E. Epler, N. F. Gardner, M. G. Craford, J. R. Wendt, J. A. Simmons, and M. M.Sigalas, “InGaN/GaN quantum-well heterostructure light-emitting diodes employing photonic crystal structures,”Appl. Phys. Lett. 84(19), 3885 (2004).1. IntroductionIII-nitride wide bandgap light-emitting diodes (LEDs) have recently attracted considerable interest due to their various applications, such as traffic signals, back-side lighting in liquid crystal display and illumination lighting by white light LEDs. Nevertheless, up to now, all commercial LEDs are grown on foreign substrates such as sapphire, GaAs, SiC and silicon due to the lack of availability of a native substrate. Such heteroepitaxial growth typically leads to high density of threading dislocations (108-1010 cm−2), which are detrimental to the LED’s internal quantum efficiency (ηint) and reliability. The development of high-brightness LEDs suffers from the non-thermal rollover of ηint at high current density, known as the efficiency droop [1,2]. To solve the above problems, the homoepitaxially grown LEDs have been presented with low dislocation density and high thermal conductivity [3–5], as a result of recent progress in production of free-standing GaN (FS-GaN) substrate grown hydride vapor phase epitaxy (HVPE). Furthermore, the availability of FS-GaN is also advantageous for LED device processing when FS-GaN can be made electrically conductive, eliminating the etching step for contact and increasing the emission areas. However, the production cost of FS-GaN substrate is still too high for a broader commercialization, which limits the further investigation of homoepitaxial LEDs. Recently, it is heart-stirring that Cich et al. has shown the good performance of GaN LEDs on bulk GaN substrates with very high external quantum efficiency (ηext) up to high current density [6].It is known that the extraction efficiency of LEDs is limited by high refractive index contrast between the GaN (2.5) and air (1.0). For conventional LED on sapphire, smaller refractive index (1.78) and higher transparency of sapphire compared to bulk GaN are favor of light extraction. In contrast, a considerable amount of downward light in homoepitaxial #209568 - $15.00 USD Received 4 Apr 2014; revised 23 May 2014; accepted 26 May 2014; published 2 Jun 2014 (C) 2014 OSA30 June 2014 | Vol. 22, No. S4 | DOI:10.1364/OE.22.0A1093 | OPTICS EXPRESS A1094LED is absorbed by the FS-GaN substrate [7]. Thus, the problems about ηextraction for the LEDs on FS-GaN are more serious than the counterpart on sapphire. Currently, it has been demonstrated that several methods are used to improve ηextraction in GaN-based LEDs on FS-GaN substrate, such as etching roughness of N-face GaN [8,9], geometric die shaping [10,11] and vertically mounted configuration [12]. The N-face GaN roughness on the backside of FS-GaN substrate may effectively increase the light extraction, but severe absorption losses also occur since some of the backside light is absorbed by mechanical support of chips due to the lack of metal reflector. In contrast, the die shaping and vertically mounted configuration open possibilities to enhance light extraction from LEDs sidewalls. However, the laser shaping process for the chips with thick bulk GaN easily causes undesirable damages to the LEDs properties. In general, until now there has still been very limited literature on how ηextraction of homoepitaxial LEDs is increased, while a majority of work is concentrated on the research of ηint and droop effect. Therefore, more efforts are needed to gain the high ηextraction for LEDs grown on FS-GaN substrate.To enhance ηextraction of LEDs, photonic crystal (PhC) structures have drawn much attention, which could lead to efficiently coupling light from the dielectric-guided modes into air [13–16]. In addition to increasing the extraction efficiency of LEDs, periodic PhC structures have the ability to enhance the directionality, especially along the vertical orientation. Recently, Weisbuch et al. demonstrated two air-gap embedded PhCs, which created a waveguide with highly confined and well-extracted mode while exhibiting no significant deleterious effects on the LEDs [17,18]. In our previous report [19], we have employed a low-cost and high-throughput method of nanospherical-lens photolithography (NLP) to fabricate two-dimensional SiO2 PhC embedded in p-GaN to improve the light extraction of LEDs on sapphire substrate. In this work, we report on the development and analysis of improved extraction efficiency of homoepitaxial LEDs on FS-GaN substrate with double SiO2 PhC structures by NLP and overgrowth. The effects of the double SiO2 PhC on the light propagation of homoepitaxial LEDs are analyzed and discussed in detail. The finite-difference time-domain (FDTD) is also used to simulate the optical field distributions to verify the experimental results.2. FabricationThe blue InGaN/GaN LEDs were grown on c-plane FS-GaN substrate with a dislocation density of about 2 × 107 cm−2 by a veeco metal-organic chemical vapor deposition (MOCVD) system at a growth pressure of 200 mbar. Figure 1 shows the fabrication process flow for embedded and double SiO2 PhC LEDs. The details of the SiO2 nanodisk structures fabricated on n-GaN had been described in our previous reports [19]. Following the SiO2 nanodisks, the regrown structure consisted of 2.5 µm n-GaN layer at 1030 °C, eight periods of In0.2GaN0.8/GaN multiple quantum wells (MQWs) at 740 °C, 40 nm thick Al0.2Ga0.8N electron-blocking layer (EBL) and a 100 nm thick p-GaN layer. Subsequently, the same SiO2 nanodisks and regrowth of p-GaN with 150 nm thickness were carried out to fulfill the double PhC LEDs. Here, the height of p-GaN and top SiO2 PhC is designed as the sameness to keep the low surface roughness of double PhC LED. For comparison, the LEDs with PhC structure embedded in n-GaN and without PhC structure were also prepared. Finally, the LEDs were fabricated with a conventional square mesa (1 × 1 mm2) using indium tin oxide (ITO) with a thickness of 200 nm as a transparent current spreading layer (TCL) and Cr/Pt/Au as the n- and p-electrodes by e-beam evaporation. To avoid the impact of light extraction, the backside of all the chips was polished and without the reflectivity mirror. The light output-current-voltage characteristics of the LEDs were measured using an Everfin-PMS50 optical spectrum analyzer and a HAAS-2000 integrating sphere at a direct current (DC) mode. Far-field radiation patterns of the LEDs were also measured in LSA 3000 LED spatial analyzer with the angular resolution of 0.1°.#209568 - $15.00 USD Received 4 Apr 2014; revised 23 May 2014; accepted 26 May 2014; published 2 Jun 2014 (C) 2014 OSA30 June 2014 | Vol. 22, No. S4 | DOI:10.1364/OE.22.0A1093 | OPTICS EXPRESS A1095Fig. 1. Schematic illustration of the process for the fabrication of embedded PhC and doublePhC LEDs.3. Results and discussionFigure 2(a) demonstrates the scanning electron microscope (SEM) image of cross-sectional view of the n-GaN laterally reovergrown over the SiO2 nanodisks in the embedded PhC LED. It is noted that the SiO2 nanodisks are fully surrounded by the GaN layer, without leaving the voids. The diameter, period and height of the embedded SiO2 nanodisk are 400, 900 and 200 nm, respectively. According to the top-view of PhC structure, there is a uniform hexagonal-lattice distribution of SiO2 nanodisks as shown in Fig. 2(b). Selectively regrowth of p-GaN with a thickness of 150 nm is carried out to fill the space between SiO2 nanodisks. Furthermore, transmission electron microscopy (TEM) is employed to investigate the crystalline quality of GaN layers that are homoepitaxially grown on the SiO2 PhC structure. As shown in Figs. 2(c) and 2(d), almost no threading dislocations can be observed in both n-GaN and MQWs structures, implying the high quality growth of GaN on FS-GaN substrate. In the epitaxial lateral overgrowth (ELOG) on sapphire, Wuu et al. reported the SiO2 array could block the dislocation propagation, but new dislocations may be introduced in the lateral coalescence region during the second growth [20]. Unlike the heteroepitaxial growth, there are also no new dislocations formed on the SiO2 nanodisks in the grown process.#209568 - $15.00 USD Received 4 Apr 2014; revised 23 May 2014; accepted 26 May 2014; published 2 Jun 2014(C) 2014 OSA30 June 2014 | Vol. 22, No. S4 | DOI:10.1364/OE.22.0A1093 | OPTICS EXPRESS A1096Fig. 2. (a) Cross-sectional view of embedded PhC of SiO 2 nanodisks surrounded by n -GaN and(b) tilted top-view of p -GaN surface with top SiO 2 PhC structure. (c) Cross-sectional TEMimage of the LED with embedded SiO 2 PhC. (d) is the magnified region of MQWs from PhCLED in (c).The logarithmic I -V curves of the LEDs without and with PhC structures are shown in Fig. 5(a). At an injection current of 350 mA, the forward bias voltages of LEDs without PhC, with an embedded PhC and double PhC are 3.57 V, 3.64 V and 3.77 V, respectively. The slight increase in forward voltage, especially for the double PhC LED due to enhanced surface roughness is acceptable for practical application. In addition, it is found that the leakage currents of these LEDs on FS-GaN without and with PhC are almost identical, about 8.2 × 10−8 A at the reverse voltage of 10 V. The reverse leakage current of InGaN/GaN has been attributed to the electron tunneling from p -GaN to n -GaN through dislocation or defects [21]. Therefore, the low and accordant reverse leakage currents of these LEDs imply comparable dislocation density of GaN for LEDs without and with SiO 2 nanodisks, consistent with the above TEM results. Furthermore, the turn-on voltage of the aforementioned LEDs is almost same, about 1.9 V as shown in the inset of Fig. 3(a), suggesting that the carrier transport characteristics are not injured by embedded SiO 2 PhC structures. As compared with the LED without PhC, the electroluminescence (EL) intensities of LED with PhC are obviously increased in Fig. 3(b). The peak wavelength of EL spectra in conventional, embedded PhC and double PhC LEDs is measured as 456.4, 455.2 and 454.1 nm at 350 mA current, respectively. The slight blue-shift phenomenon in the EL spectra is caused by a partial compression strain release in the InGaN active layer through selective regrowth on SiO 2 nanodisks. Figure 3(c) shows the light output power as functions of current for the PhC LEDs. At 350 mA current, the light output power of LEDs with embedded PhC and double PhC is increased by 29.9% and 47.2%, respectively, compared to that without PhC. Based on the observed increases of 47.2% for double PhC and 29.9% for embedded PhC, the increase of ηextraction is estimated to be only 13.3% for top SiO 2 PhC due to diffused scattering effect. In fact, light extraction of surface PhC is to do rather more than that. The overlap of photon escape probability from the surface PhC and embedded PhC structures weakens the contribution of double SiO 2 PhC.#209568 - $15.00 USDReceived 4 Apr 2014; revised 23 May 2014; accepted 26 May 2014; published 2 Jun 2014(C) 2014 OSA 30 June 2014 | Vol. 22, No. S4 | DOI:10.1364/OE.22.0A1093 | OPTICS EXPRESS A1097Fig. 3. (a) The reverse and forward I-V characteristics of the LEDs without and with PhCstructures on a logarithmic scale. Inset: corresponding enlarged forward I-V curves closed tothe turn-on voltage. (b) Peak wavelength of the EL emission spectra of the LEDs at a current of350 mA. (c) Optical output power (L-I) of these LEDs as a function of injection current.We measure the light output radiation patterns of the LEDs with and without PhC structures at a driving current of 200 mA. Here, the chips are Au-wire bonded and loaded on an aluminium leaded chip carrier without epoxy encapsulation. The far-field emission patterns from the PhC LEDs show much smaller view angles but obvious enhancement in the overall integrated emission intensity. In order to scrutinize the effect of the PhCs, the normalized relative radiation profiles of different LEDs are plotted in Fig. 4. It is found that the full-width-at-half maximum (FWHM) of emission divergence for the embedded and double PhC LEDs are 128.1° and 131.6°, respectively, compared to that of 159.4° without SiO2 PhC structure. Here, the errors of the measured divergence values are ± 0.1°. In general, the conventional LED on sapphire substrate has a divergent angle of about 151° on the same LED structure [19]. Here, the larger divergent angle in LED without PhC on FS-GaN substrate implies that the light confined in the LED chip is extracted from edge of the chip or of the FS-GaN substrate after multiple scattering or reflection [22], corresponding to the low light extraction efficiency (LEE). In the PhC LEDs, the light beam shaping of SiO2 nanodisks is highly remarkable with a view angle reduction of about 30°, which helps to confine the light to radiate in the vertical direction for the suppression of total internal reflection. Especially, in the double PhC LED, the light is further redirected to the top escape cone through the twice transmission of embedded and top SiO2 arrays, resulting in the most significant focusing effect and more photon capable of escaping from the chips.#209568 - $15.00 USD Received 4 Apr 2014; revised 23 May 2014; accepted 26 May 2014; published 2 Jun 2014 (C) 2014 OSA30 June 2014 | Vol. 22, No. S4 | DOI:10.1364/OE.22.0A1093 | OPTICS EXPRESS A1098Fig. 4. Normalized relative far-field emission patterns of the LEDs without and with SiO2 PhCstructures at a driving current of 200 mA.To further understand the influence of embedded PhC and double PhC patterns on the performance of LEDs, a two-dimensional (2D) finite difference time domain (FDTD) simulation is used. Here, considering the plane isotropy of hexagonal PhC structures, the simulation model is simplified from 3D to 2D to reduce the calculation time. We use the perfectly matched layer (PML) boundary condition for the simulation and a point dipole polarized along the x, y and z directions is used as radiation source. The computational domains are 42 µm in the x-direction and 16 µm in the y-direction. The simulated LED structure consists of 250 nm thick p-GaN/150 nm thick MQWs/4.2 µm thick n-GaN/10 µm FS-GaN substrate, as shown in Fig. 5(a). The shape and size of the SiO2 PhC solid model for FDTD simulation are determined are exactly the same as those in the SEM images in Fig. 2 and scarce non-uniformities in PhC are ignored to simplify the simulation. Figures 5(b)-5(d) compare the propagation of electro-magnetic waves passing through SiO2 PhC structures and hybrid surface. The sole GaN-air interface produces the torch-like radiation pattern for LED without PhC on FS-GaN, resulting in the stronger electric field distribution in the FS-GaN and lower LEE. In contrast, part of optical energy for LED on sapphire is radiated to air through light incidence at the mesa sidewall due to the diffraction of interface between GaN and sapphire [19]. In the PhC LEDs on FS-GaN, the incorporation of SiO2 nanodisks is seen to suppress the internal reflection and diffract the optical energy to the surface normal. Due to the suited dimension of SiO2 with wavelength, wave-like features such as interference and diffraction dominate the interface between GaN and air. Especially, the PhC converging effect of embedded and top SiO2 structures in double PhC LED is coupled and causes the smallest view angle, consistent with the observation far-field emission patterns in Fig. 4.#209568 - $15.00 USD Received 4 Apr 2014; revised 23 May 2014; accepted 26 May 2014; published 2 Jun 2014(C) 2014 OSA30 June 2014 | Vol. 22, No. S4 | DOI:10.1364/OE.22.0A1093 | OPTICS EXPRESS A1099Fig. 5. (a) The schematic representation of the 2D FDTD simulation for the double PhC LED.FDTD simulation of light propagation in (b) LED without PhC, (c) LED with embedded PhCand (d) LED with double PhC on FS-GaN substrate.4. ConclusionIn summary, we fabricate the SiO2 nanodisk arrays to form PhC structures, which are embedded in n-GaN and p-GaN of homoepitaxially grown LEDs by NLP and overgrowth. Significant improvements on light extraction of the embedded PhC and double PhC LED at 350 mA current have been observed of up to 29.9% and 47.2% over that without PhC structure, respectively. Furthermore, the view angle of PhC LEDs is obviously reduced, which confines the light to radiate in the vertical direction for the suppression of total internal reflection. The work offers a promising potential to reduce the severe internal reflection and increase the light output powers of homoepitaxial LEDs on FS-GaN substrate. AcknowledgmentThis work was supported by the National Natural Sciences Foundation of China under Grant 61274040, 61274008 and 51102226, the National Basic Research Program of China under Grant 2011CB301902, the National High Technology Program of China under Grant 2014AA032605 and Youth Innovation Promotion Association, Chinese Academy of Sciences.#209568 - $15.00 USD Received 4 Apr 2014; revised 23 May 2014; accepted 26 May 2014; published 2 Jun 2014(C) 2014 OSA30 June 2014 | Vol. 22, No. S4 | DOI:10.1364/OE.22.0A1093 | OPTICS EXPRESS A1100。