计算化学公社

标题: oniom 计算报错Linear angle in Tors. L402 [打印本页]

作者
Author: 晓来雨过 时间: 2021-3-25 13:18
标题: oniom 计算报错Linear angle in Tors. L402
各位老师好，我想通过oniom来模拟晶体环境中优化分子的激发态，但是会出现 Linear angle in Tors. Error termination via Lnk1e in /home/xy/g09/l402.exe的报错，搜索后，并没有发现L402报错的解决方案。请问这样的问题如何解决？输入输出文件如下，谢谢老师！

作者
Author: sobereva 时间: 2021-3-25 16:57
改成opt(nomicro,cartesian)再试

注意用QEq电荷完全不足以合理描述周围环境对高层分子的静电作用，此文说了
基于背景电荷计算分子在晶体环境中的吸收光谱
http://sobereva.com/579（http://bbs.keinsci.com/thread-20960-1-1.html）

作者
Author: 晓来雨过 时间: 2021-3-27 21:39

sobereva 发表于 2021-3-25 16:57
改成opt(nomicro,cartesian)再试

注意用QEq电荷完全不足以合理描述周围环境对高层分子的静电作用，此文 ...

老师，按照您说的仍然报错。您在这个帖子里说“只有当涉及到几何优化的问题，比如在晶体环境中优化分子的激发态，那才真的有必要用ONIOM，因为这样才能在优化时表现必不可少的环境分子和中心分子间的范德华作用”。我恰好就是想利用oniom来计算该体系在固态下的荧光发射。

作者
Author: sobereva 时间: 2021-3-28 10:27

晓来雨过发表于 2021-3-27 21:39
老师，按照您说的仍然报错。您在这个帖子里说“只有当涉及到几何优化的问题，比如在晶体环境中优化分子的 ...

上传压缩后的用了opt(nomicro,cartesian)的输出文件

文中强调了恰当设置电荷的重要性，所以你用QEq根本不行

作者
Author: 晓来雨过 时间: 2021-3-28 12:29
本帖最后由晓来雨过于 2021-3-28 16:00 编辑

sobereva 发表于 2021-3-28 10:27
上传压缩后的用了opt(nomicro,cartesian)的输出文件

文中强调了恰当设置电荷的重要性，所以你用QEq根 ...

老师，该out文件即采用了opt(nomicro,cartesian)，您提及的设置电荷问题我会及时更正。

作者
Author: sobereva 时间: 2021-3-28 23:13

晓来雨过发表于 2021-3-28 12:29
老师，该out文件即采用了opt(nomicro,cartesian)，您提及的设置电荷问题我会及时更正。

不好说。貌似要求用cartesian做优化的设置没生效。建议发邮件到官方邮箱问问，可能需要特殊的做法

作者
Author: 晓来雨过 时间: 2021-3-28 23:31

sobereva 发表于 2021-3-28 23:13
不好说。貌似要求用cartesian做优化的设置没生效。建议发邮件到官方邮箱问问，可能需要特殊的做法

明白，谢谢老师！

作者
Author: 桌坛新星10500 时间: 2021-3-29 11:03

晓来雨过发表于 2021-3-28 23:31
明白，谢谢老师！

老师您好，我在处理类似的体系，荧光分子的oniom模型模拟晶体环境中优化激发态是报了和您一样的错误。老师您可否愿意分享您的进展和心得，万分感激。

作者
Author: 晓来雨过 时间: 2021-3-30 16:39

桌坛新星10500 发表于 2021-3-29 11:03
老师您好，我在处理类似的体系，荧光分子的oniom模型模拟晶体环境中优化激发态是报了和您一样的错误。老 ...

官方给的邮件如下，我还没有尝试，咋感觉是在向我推销GV6~~
The problem comes from the initial geometry, including the connectivity and atom types. For Molecular Mechanics methods, atom types and bond connectivities are very important. However, in the input file, all bonds are labeled as single bonds, which means that the atom types are likely incorrect either.

You may open the initial geometry, go to "Edit" -> "Rebond", so that the bonds are not just single bonds, then add the Cd-O, Cd-N bonds and O-H hydrogen bonds. If you have Gaussview 6, it has a new feature that you can change the default criteria to identify the bonds, so you can speed up this task. If you have GaussView 5, then you'll have to do it one by one.

Open Gaussview 6, go to "Tools" -> "Custom Bonding Parameters...", click "New", choose element 1 (Elem.1) and 2 as O and Cd, respectively, load the "Default Lengths" first, then type a custom value as the upper bound for O-Cd (e.g. 3) under "weak", so that Gaussview 6 can identify all the O-Cd weak bonds with the new cutoff. Similarly you can add N-Cd bonds and O-H hydrogen bonds. Then it is a lot easier to do further adjustments one by one. An example input file with edited connectivity is attached.

After editing the connectivity, you can obtain a better initial geometry by hitting the "clean" button (the one with a broom).

To fix the atom type, go to the "Atom List Editor" in GaussView ("Tools" -> "Atom List..."), click on the "M" button ("Show MM Type Columns"). If they don't seem right, go to "Edit" -> "Update MM Types and Charges", and choose "All Atoms" or "Selected Atoms" according to your chemical understanding of the system.

Finally, electronic embedding are more expensive than mechanical embedding, and convergence are normally harder too. For ONIOM TD frequency takes much more memory than mechanical embedding, so you may do the opt+freq with mechanical embedding followed by a single-point with electronic embedding, instead of TD Opt Freq with electronic embedding; or use the mechanical embedding for pre-optimization, followed by re-optimization with electronic embedding.

作者
Author: 桌坛新星10500 时间: 2021-3-31 09:29

晓来雨过发表于 2021-3-30 16:39
官方给的邮件如下，我还没有尝试，咋感觉是在向我推销GV6~~
The problem comes from the initial geomet ...

您的消息对我非常有启发，谢谢老师。

作者
Author: beefly 时间: 2021-3-31 10:17
oniom内部已经绑定了冗余内坐标优化，opt(cartesian)不起作用。可能这是个bug

欢迎光临计算化学公社 (http://bbs.keinsci.com/)