Hi Partha,

Thank you for your question.=

Atom-in-material properties should not explicitly= depend on the basis set used to perform the quantum chemistry calculation.= The reason for this is that nature corresponds to the limit of an arbitrar= ily large basis set (aka 'complete basis set limit'). For reasons o= f computational efficiency, in practical calculations a finite basis set is= usually used to get an answer quickly. So, we want to use population analy= sis methods that are not too sensitive to the choice of basis set, so that = reasonably sized basis sets will produce answers that are not much differen= t than what would be computed if an arbitrarily large basis set (aka 'c= omplete basis set limit') were used.=C2=A0

The= Mulliken method explicitly depends on the basis set choice and has no comp= lete basis set limit. In other words, the Mulliken populations are not defi= ned for arbitrarily large basis set. This is very bad, because it means tha= t as the accuracy/precision of the quantum chemistry calculation is improve= d by using larger and larger basis set, the Mulliken charges get worse and = worse with no mathematical limits or meaning. So, you should not use Mullik= en populations. It also means that if you run a similar calculation twice, = but using different basis sets for each calculation, the Mulliken charges t= hat you get from the two similar calculations may not be similar. For examp= le, Mulliken population analysis may predict the cations have turned into a= nions, and vice versa, even though it is exactly the same material. In othe= r words, Mulliken population analysis often gets cations and anions confuse= d; it does not know the distinction between them.

= Between the two methods you mentioned, natural population analysis (NPA) is= the better choice. An even better choice would be to use a method like DDE= C6 for which the atom-in-material properties are a functional of the electr= on and spin density distributions with no explicit basis set dependence, an= d for which the atom-in-material properties are chemically meaningful acros= s a broad range of material types.

Sincerely,

Tom

On Sun, Jun 28, 2020 at 8:43 A= M Partha Sengupta anapspsmo%x%gmail.com &l= t;owner-chemistry%x%ccl.net>= ; wrote:

Sir,=C2=A0For a metal complexes involving Cu[N,O donor], [Chlorine and fluorine atoms are= in the benzene ring]. =C2=A0I found that cu has 0.96091 =C2=A0charges( Natural Population analysis) while Mulliken atomic =C2=A0charge is 0.457333. On the same process N has -0.53717 and -0.29941 respectively. = What is the reason behind this? What will be the better choice to represent?
Part= ha Sengupta

--
Dr. Partha Sarathi Sengupta
Associate Professor=
Vivekananda Mahavidyalaya, Burdwan

--0000000000000712e605a927d49c-- From owner-chemistry@ccl.net Sun Jun 28 14:30:00 2020 From: "Peeter Burk peeter.burk..ut.ee" To: CCL Subject: CCL: Charge Message-Id: <-54117-200628142731-6921-ZaKYCAjQnZRPW0BhTZ6iSA#server.ccl.net> X-Original-From: Peeter Burk Content-Language: en-US Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=utf-8; format=flowed Date: Sun, 28 Jun 2020 21:27:20 +0300 MIME-Version: 1.0 Sent to CCL by: Peeter Burk [peeter.burk-.-ut.ee] Hi, Can we even discuss which charge partitioning scheme is better (aside of convergency with increasing basis set size) as AFAIK atomic charges are not a measurable property of atoms in molecules or materials? Chemists (including myself) like to use them to explain properties like reactivity, but as long as you use them within series most schemes give the similar answers (if you compare trends and changes, not absolute values)... Best regards Peeter On 28/06/2020 19:56, Thomas Manz thomasamanz(a)gmail.com wrote: > Hi Partha, > > Thank you for your question. > > Atom-in-material properties should not explicitly depend on the basis > set used to perform the quantum chemistry calculation. The reason for > this is that nature corresponds to the limit of an arbitrarily large > basis set (aka 'complete basis set limit'). For reasons of > computational efficiency, in practical calculations a finite basis set > is usually used to get an answer quickly. So, we want to use > population analysis methods that are not too sensitive to the choice > of basis set, so that reasonably sized basis sets will produce answers > that are not much different than what would be computed if an > arbitrarily large basis set (aka 'complete basis set limit') were used. > > The Mulliken method explicitly depends on the basis set choice and has > no complete basis set limit. In other words, the Mulliken populations > are not defined for arbitrarily large basis set. This is very bad, > because it means that as the accuracy/precision of the quantum > chemistry calculation is improved by using larger and larger basis > set, the Mulliken charges get worse and worse with no mathematical > limits or meaning. So, you should not use Mulliken populations. It > also means that if you run a similar calculation twice, but using > different basis sets for each calculation, the Mulliken charges that > you get from the two similar calculations may not be similar. For > example, Mulliken population analysis may predict the cations have > turned into anions, and vice versa, even though it is exactly the same > material. In other words, Mulliken population analysis often gets > cations and anions confused; it does not know the distinction between > them. > > Between the two methods you mentioned, natural population analysis > (NPA) is the better choice. An even better choice would be to use a > method like DDEC6 for which the atom-in-material properties are a > functional of the electron and spin density distributions with no > explicit basis set dependence, and for which the atom-in-material > properties are chemically meaningful across a broad range of material > types. > > Sincerely, > > Tom > > > On Sun, Jun 28, 2020 at 8:43 AM Partha Sengupta anapspsmo%x%gmail.com >

> wrote: > > Sir, For a metal complexes involving Cu[N,O donor], [Chlorine and > fluorine atoms are in the benzene ring]. I found that cu has > 0.96091 charges( Natural Population analysis) while Mulliken > atomic charge is 0.457333. On the same process N has -0.53717 and > -0.29941 respectively. What is the reason behind this? What will > be the better choice to represent? > Partha Sengupta > > -- > */Dr. Partha Sarathi Sengupta > Associate Professor > Vivekananda Mahavidyalaya, Burdwan/* > From owner-chemistry@ccl.net Sun Jun 28 17:36:01 2020 From: "Robert Molt r.molt.chemical.physics%a%gmail.com" To: CCL Subject: CCL: Charge Message-Id: <-54118-200628173530-27108-htjth6KDOE13Obn5B+J3vg_+_server.ccl.net> X-Original-From: Robert Molt Content-Type: multipart/alternative; boundary="Apple-Mail=_892DEB13-3261-4D31-9313-85CCF68387FB" Date: Sun, 28 Jun 2020 17:35:21 -0400 Mime-Version: 1.0 (Mac OS X Mail 12.4 \(3445.104.11\)) Sent to CCL by: Robert Molt [r.molt.chemical.physics/./gmail.com] --Apple-Mail=_892DEB13-3261-4D31-9313-85CCF68387FB Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset=utf-8 People can have have discussions of the best charge model as they see = fit. I make no comment on comparative charge models. I do offer the reminder that atomic charge is not an observable. This is = a construct in the minds of chemists. It is literally impossible to do = an experiment to measure the notion. One of the many reasons for this is = the fact that one cannot reduce the electron density (a function of 3 = spatial dimensions, at least) to being =E2=80=9Cowned=E2=80=9D by one = atom, as opposed to distributed to unequal degrees in space. I concur with Dr. Burk=E2=80=99s claim that atomic charges are, at best, = an internal scheme. If you see inconsistency across charge schemes, it = is not the fault of a given scheme, but the notion itself. > On Jun 28, 2020, at 2:27 PM, Peeter Burk peeter.burk..ut.ee = wrote: >=20 >=20 > Sent to CCL by: Peeter Burk [peeter.burk-.-ut.ee] > Hi, >=20 > Can we even discuss which charge partitioning scheme is better (aside = of convergency with increasing basis set size) as AFAIK atomic charges = are not a measurable property of atoms in molecules or materials? >=20 > Chemists (including myself) like to use them to explain properties = like reactivity, but as long as you use them within series most schemes = give the similar answers (if you compare trends and changes, not = absolute values)... >=20 > Best regards >=20 > Peeter >=20 > On 28/06/2020 19:56, Thomas Manz thomasamanz(a)gmail.com wrote: >> Hi Partha, >>=20 >> Thank you for your question. >>=20 >> Atom-in-material properties should not explicitly depend on the basis = set used to perform the quantum chemistry calculation. The reason for = this is that nature corresponds to the limit of an arbitrarily large = basis set (aka 'complete basis set limit'). For reasons of computational = efficiency, in practical calculations a finite basis set is usually used = to get an answer quickly. So, we want to use population analysis methods = that are not too sensitive to the choice of basis set, so that = reasonably sized basis sets will produce answers that are not much = different than what would be computed if an arbitrarily large basis set = (aka 'complete basis set limit') were used. >>=20 >> The Mulliken method explicitly depends on the basis set choice and = has no complete basis set limit. In other words, the Mulliken = populations are not defined for arbitrarily large basis set. This is = very bad, because it means that as the accuracy/precision of the quantum = chemistry calculation is improved by using larger and larger basis set, = the Mulliken charges get worse and worse with no mathematical limits or = meaning. So, you should not use Mulliken populations. It also means that = if you run a similar calculation twice, but using different basis sets = for each calculation, the Mulliken charges that you get from the two = similar calculations may not be similar. For example, Mulliken = population analysis may predict the cations have turned into anions, and = vice versa, even though it is exactly the same material. In other words, = Mulliken population analysis often gets cations and anions confused; it = does not know the distinction between them. >>=20 >> Between the two methods you mentioned, natural population analysis = (NPA) is the better choice. An even better choice would be to use a = method like DDEC6 for which the atom-in-material properties are a = functional of the electron and spin density distributions with no = explicit basis set dependence, and for which the atom-in-material = properties are chemically meaningful across a broad range of material = types. >>=20 >> Sincerely, >>=20 >> Tom >>=20 >>=20 >> On Sun, Jun 28, 2020 at 8:43 AM Partha Sengupta anapspsmo%x%gmail.com =

> wrote: >>=20 >> Sir, For a metal complexes involving Cu[N,O donor], [Chlorine and >> fluorine atoms are in the benzene ring]. I found that cu has >> 0.96091 charges( Natural Population analysis) while Mulliken >> atomic charge is 0.457333. On the same process N has -0.53717 and >> -0.29941 respectively. What is the reason behind this? What will >> be the better choice to represent? >> Partha Sengupta >>=20 >> -- */Dr. Partha Sarathi Sengupta >> Associate Professor >> Vivekananda Mahavidyalaya, Burdwan/* >>=20 >=20 >=20 >=20 > -=3D This is automatically added to each message by the mailing script = =3D- > To recover the email address of the author of the message, please = change>=20>=20>=20=>=20>=20Conferences: = http://server.ccl.net/chemistry/announcements/conferences/ >=20>=20>=20>=20 >=20 --Apple-Mail=_892DEB13-3261-4D31-9313-85CCF68387FB Content-Transfer-Encoding: quoted-printable Content-Type: text/html; charset=utf-8 People can have have discussions of the best charge model as = they see fit. I make no comment on comparative charge models.

I do offer the reminder = that atomic charge is not an observable. This is a = construct in the minds of chemists. It is literally impossible to do an = experiment to measure the notion. One of the many reasons for this is = the fact that one cannot reduce the electron density (a function of 3 = spatial dimensions, at least) to being =E2=80=9Cowned=E2=80=9D by one = atom, as opposed to distributed to unequal degrees in space.

I concur with Dr. = Burk=E2=80=99s claim that atomic charges are, at best, an internal = scheme. If you see inconsistency across charge schemes, it is not the = fault of a given scheme, but the notion itself.

On Jun 28, 2020, at 2:27 PM, Peeter Burk peeter.burk..ut.ee = <owner-chemistry!A!ccl.net> wrote:

Sent to CCL by: Peeter Burk [peeter.burk-.-ut.ee]
Hi,

Can we even discuss which = charge partitioning scheme is better (aside of convergency with = increasing basis set size) as AFAIK atomic charges are not a measurable = property of atoms in molecules or materials?

Chemists (including myself) like to use them to explain = properties like reactivity, but as long as you use them within series = most schemes give the similar answers (if you compare trends and = changes, not absolute values)...

Best = regards

Peeter

On 28/06/2020 19:56, Thomas Manz thomasamanz(a)gmail.com wrote:
Hi Partha,

Thank you for your question.

Atom-in-material properties should not explicitly depend on = the basis set used to perform the quantum chemistry calculation. The = reason for this is that nature corresponds to the limit of an = arbitrarily large basis set (aka 'complete basis set limit'). For = reasons of computational efficiency, in practical calculations a finite = basis set is usually used to get an answer quickly. So, we want to use = population analysis methods that are not too sensitive to the choice of = basis set, so that reasonably sized basis sets will produce answers that = are not much different than what would be computed if an arbitrarily = large basis set (aka 'complete basis set limit') were used.

The Mulliken method explicitly depends on the = basis set choice and has no complete basis set limit. In other words, = the Mulliken populations are not defined for arbitrarily large basis = set. This is very bad, because it means that as the accuracy/precision = of the quantum chemistry calculation is improved by using larger and = larger basis set, the Mulliken charges get worse and worse with no = mathematical limits or meaning. So, you should not use Mulliken = populations. It also means that if you run a similar calculation twice, = but using different basis sets for each calculation, the Mulliken = charges that you get from the two similar calculations may not be = similar. For example, Mulliken population analysis may predict the = cations have turned into anions, and vice versa, even though it is = exactly the same material. In other words, Mulliken population analysis = often gets cations and anions confused; it does not know the distinction = between them.

Between the two methods you = mentioned, natural population analysis (NPA) is the better choice. An = even better choice would be to use a method like DDEC6 for which the = atom-in-material properties are a functional of the electron and spin = density distributions with no explicit basis set dependence, and for = which the atom-in-material properties are chemically meaningful across a = broad range of material types.

Sincerely,

Tom

On Sun, Jun 28, 2020 at 8:43 AM Partha Sengupta = anapspsmo%x%gmail.com <http://gmail.com> = <owner-chemistry*_*ccl.net = <mailto:owner-chemistry*_*ccl.net>> wrote:

=    Sir, For a metal complexes involving Cu[N,O donor], = [Chlorine and
   fluorine atoms are in the = benzene ring]. I found that cu has
=    0.96091 charges( Natural Population analysis) while = Mulliken
   atomic charge is 0.457333. On = the same process N has -0.53717 and
=    -0.29941 respectively. What is the reason behind this? = What will
   be the better choice to = represent?
   Partha Sengupta

   -- =     */Dr. Partha Sarathi Sengupta
=    Associate Professor
=    Vivekananda Mahavidyalaya, Burdwan/*

-=3D = This is automatically added to each message by the mailing script =3D-
To recover the email address of the author of the message, = please change
the strange characters on the top line to = the !A! sign. You can also
look up the X-Original-From: line = in the mail header.

E-mail to subscribers: = CHEMISTRY!A!ccl.net or = use:
    http://www.ccl.net/cgi-bin/ccl/send_ccl_message

E-mail to administrators: CHEMISTRY-REQUEST!A!ccl.net or use
=     http://www.ccl.net/cgi-bin/ccl/send_ccl_message

http://www.ccl.net/chemistry/sub_unsub.shtml

Before posting, check wait time at: http://www.ccl.net

Job: http://www.ccl.net/jobs Conferences: http://server.ccl.net/chemistry/announcements/conferences/<= br class=3D"">
Search Messages: http://www.ccl.net/chemistry/searchccl/index.shtml

If your mail bounces from CCL with 5.7.1 = error, check:
    http://www.ccl.net/spammers.txt

RTFI: http://www.ccl.net/chemistry/aboutccl/instructions/

= --Apple-Mail=_892DEB13-3261-4D31-9313-85CCF68387FB-- From owner-chemistry@ccl.net Sun Jun 28 18:27:00 2020 From: "Thomas Manz thomasamanz|-|gmail.com" To: CCL Subject: CCL: Charge Message-Id: <-54119-200628181448-9171-wlC+xf0E2JVOpE7UF0Tl4g]=[server.ccl.net> X-Original-From: Thomas Manz Content-Type: multipart/alternative; boundary="000000000000bfdce405a92c442b" Date: Sun, 28 Jun 2020 16:14:29 -0600 MIME-Version: 1.0 Sent to CCL by: Thomas Manz [thomasamanz__gmail.com] --000000000000bfdce405a92c442b Content-Type: text/plain; charset="UTF-8" Hi Peeter, >Can we even discuss which charge partitioning scheme is better (aside of convergency with increasing basis set size) as AFAIK atomic charges are not a measurable property of atoms in molecules or materials? Two things worth considering: (A) Even though the charges of atoms in materials are not directly observable, it is possible to examine the extent of correlations between atomic charges and experimentally observable properties. The better methods for computing net atomic charges give values that are well correlated to a wide array of experimentally observable properties across diverse material types. What is extremely interesting is that a charge assignment method can be constructed that exhibits average or better correlations to each and every chemical and physical property related to atomic charges. Construction of such a method is not necessarily unique, but two different charge assignment schemes that simultaneously achieve this property will also be highly correlated to each other. The analogy I like to use for this is a group of darts thrown towards a target. Compared to the other darts in the group, a centrally located dart will land at least average or closer to any conceivable target. This is true even for hypothetical targets or targets not yet discovered. For example, if a new physical property related to net atomic charges is to be discovered tomorrow, we would still find this newly discovered property to be well-correlated to the net atomic charge values of the centrally located charge assignment method, even though the physical property was discovered after the charge assignment method was already developed. It is by virtue of its centrality that such a method can perform average or better across all conceivable target applications. >as long as you use them within series most schemes give the similar answers (if you compare trends and changes, not absolute values) This is true at least for molecular systems, where several different methods work at least sometimes, and moreover the centrally located method (analogous to the centrally located dart) exhibits the best overall correlations to atomic charge-related physical and chemical properties. A centrally located method has the highest overall correlations to all of the other charge assignment methods for molecular systems. I have been working for the past few months on a journal manuscript on this very topic, which I hope to submit within a week. The consistency of correlation is actually amazing. Out of 21 charge assignment methods with a complete basis set limit, the centrally located method exhibits a correlation coefficient >= 0.9 to 16 different charge assignment methods for molecules. So, in answer to your original question, we can say that a centrally located method exhibits performance that is optimal for general-purpose use across molecules. For dense solids, only a small number of charge assignment methods developed to date give reasonable results. Encouragingly, one of the centrally located methods for molecular systems is also a top performer for dense solids. (B) It is useful to view atom-in-material properties holistically rather than only in isolation. In other words, the extent of chemical consistency between different atom-in-material descriptors should be considered when choosing an atomic population analysis method: net atomic charges, atomic spin moments, bond orders, s-p-d-f-g populations (and orbital hybridization states), bond order components, polarizabilities, and a host of other atom-in-material properties. A good atomic population analysis method should give chemically consistent results across all these atom-in-material properties. Systems can be carefully chosen to test this. For example, X,C60 endohedral complexes have natural relationships between the charge and spin transferred to the C60 cage from the central atom; this can be used to test whether the net atomic charges and atomic spin moments assigned by an atomic population analysis method are chemically consistent. Encouragingly, a top performer across this criterion is also a top performer across the molecular systems mentioned in (a) above and is also a top performer for dense solids. So, in response to your original question, it turns out that certain methods consistently perform very well across multiple different criteria, which suggests there really are some outstanding atomic population analysis method(s) that are well-suited for broad use. Sincerely, Tom On Sun, Jun 28, 2020 at 1:45 PM Peeter Burk peeter.burk..ut.ee < owner-chemistry,ccl.net> wrote: > > Sent to CCL by: Peeter Burk [peeter.burk-.-ut.ee] > Hi, > > Can we even discuss which charge partitioning scheme is better (aside of > convergency with increasing basis set size) as AFAIK atomic charges are > not a measurable property of atoms in molecules or materials? > > Chemists (including myself) like to use them to explain properties like > reactivity, but as long as you use them within series most schemes give > the similar answers (if you compare trends and changes, not absolute > values)... > > Best regards > > Peeter > > On 28/06/2020 19:56, Thomas Manz thomasamanz(a)gmail.com wrote: > > Hi Partha, > > > > Thank you for your question. > > > > Atom-in-material properties should not explicitly depend on the basis > > set used to perform the quantum chemistry calculation. The reason for > > this is that nature corresponds to the limit of an arbitrarily large > > basis set (aka 'complete basis set limit'). For reasons of > > computational efficiency, in practical calculations a finite basis set > > is usually used to get an answer quickly. So, we want to use > > population analysis methods that are not too sensitive to the choice > > of basis set, so that reasonably sized basis sets will produce answers > > that are not much different than what would be computed if an > > arbitrarily large basis set (aka 'complete basis set limit') were used. > > > > The Mulliken method explicitly depends on the basis set choice and has > > no complete basis set limit. In other words, the Mulliken populations > > are not defined for arbitrarily large basis set. This is very bad, > > because it means that as the accuracy/precision of the quantum > > chemistry calculation is improved by using larger and larger basis > > set, the Mulliken charges get worse and worse with no mathematical > > limits or meaning. So, you should not use Mulliken populations. It > > also means that if you run a similar calculation twice, but using > > different basis sets for each calculation, the Mulliken charges that > > you get from the two similar calculations may not be similar. For > > example, Mulliken population analysis may predict the cations have > > turned into anions, and vice versa, even though it is exactly the same > > material. In other words, Mulliken population analysis often gets > > cations and anions confused; it does not know the distinction between > > them. > > > > Between the two methods you mentioned, natural population analysis > > (NPA) is the better choice. An even better choice would be to use a > > method like DDEC6 for which the atom-in-material properties are a > > functional of the electron and spin density distributions with no > > explicit basis set dependence, and for which the atom-in-material > > properties are chemically meaningful across a broad range of material > > types. > > > > Sincerely, > > > > Tom > > > > > > On Sun, Jun 28, 2020 at 8:43 AM Partha Sengupta anapspsmo%x%gmail.com > >

> > wrote: > > > > Sir, For a metal complexes involving Cu[N,O donor], [Chlorine and > > fluorine atoms are in the benzene ring]. I found that cu has > > 0.96091 charges( Natural Population analysis) while Mulliken > > atomic charge is 0.457333. On the same process N has -0.53717 and > > -0.29941 respectively. What is the reason behind this? What will > > be the better choice to represent? > > Partha Sengupta > > > > -- > > */Dr. Partha Sarathi Sengupta > > Associate Professor > > Vivekananda Mahavidyalaya, Burdwan/*> > > --000000000000bfdce405a92c442b Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable

Hi Peeter,

>Can we = even discuss which charge partitioning scheme is better (aside of convergen= cy with increasing basis set size) as AFAIK atomic charges are not a measur= able property of atoms in molecules or materials?=C2=A0

Two things worth considering:

(A) Even thoug= h the charges of atoms in materials are not directly observable, it is poss= ible to examine the extent of correlations between atomic charges and exper= imentally observable properties. The better methods for computing net atomi= c charges give values that are well correlated to a wide array of experimen= tally observable properties across diverse material types.

What i= s extremely interesting is that a charge assignment method=C2=A0can be cons= tructed that exhibits average or better correlations to each and every chem= ical and physical property related to atomic charges. Construction of such = a method is not necessarily unique, but two different charge assignment sch= emes that simultaneously achieve this property will also be highly correlat= ed to each other. The analogy I like to use for this is a group of darts th= rown towards a target. Compared to the other darts in the group, a=C2=A0cen= trally located dart will land at least average or closer to any conceivable= target. This is true even for hypothetical targets or targets not yet disc= overed. For example, if a new physical property related to net atomic charg= es is to be discovered tomorrow, we would still find this newly discovered = property to be well-correlated to the net atomic charge values of the centr= ally located charge assignment method, even though the physical property wa= s discovered after the charge assignment method was already developed. It i= s by virtue of its centrality that such a method can perform average or bet= ter across all conceivable target applications.

&g= t;as long as you use them within series most schemes give the similar answe= rs (if you compare trends and changes, not absolute values)

<= /div>

This is true at least for molecular systems, where several differ= ent methods work at least sometimes, and moreover the centrally located met= hod (analogous to the centrally located dart) exhibits the best overall cor= relations to atomic charge-related physical and chemical properties. A cent= rally located method has the highest overall correlations to all of the oth= er charge assignment methods for molecular systems. I have been working for= the past few months on a journal manuscript on this very topic, which I ho= pe to submit within a week. The consistency of correlation is actually amaz= ing. Out of 21 charge assignment methods with a complete basis set limit, t= he centrally located method exhibits a correlation coefficient >=3D 0.9 = to 16 different charge assignment methods for molecules.=C2=A0

So, in answer to your original question, we can say that a cen= trally located method exhibits performance that is optimal for general-purp= ose use across molecules. For dense solids, only a small number of charge a= ssignment methods developed to date give reasonable results. Encouragingly,= one of the=C2=A0centrally located methods for molecular systems is also a = top performer for dense solids.

(B) It is useful t= o view atom-in-material properties holistically rather than only in isolati= on. In other words, the extent of chemical consistency between different at= om-in-material descriptors should be considered when choosing an atomic pop= ulation analysis method: net atomic charges, atomic spin moments, bond orde= rs, s-p-d-f-g populations (and orbital hybridization states), bond order co= mponents, polarizabilities, and a host of other atom-in-material properties= . A good atomic population analysis method should give chemically consisten= t results across all these atom-in-material properties. Systems can be care= fully chosen to test this. For example, X,C60 endohedral complexes have nat= ural relationships between the charge and spin transferred to the C60 cage = > from the central atom; this can be used to test whether the net atomic char= ges and atomic spin moments=20 assigned=20 by an atomic population analysis method are chemically consistent. Encourag= ingly, a top performer across this criterion is also a top performer across= the molecular systems mentioned in (a) above and is also a top performer f= or dense solids.

So, in response to your original = question, it turns out that certain methods consistently perform very well = across multiple different criteria, which suggests there really are some ou= tstanding atomic population analysis method(s) that are well-suited for bro= ad use.=C2=A0

Sincerely,

= Tom

On Sun, Jun 28, 2020 at 1:45 PM Peeter Burk peeter.burk..ut.ee <= owner-chemistry,ccl.net> wrote:

Sent to CCL by: Peeter Burk [peeter.burk-.-ut.ee]
Hi,

Can we even discuss which charge partitioning scheme is better (aside of convergency with increasing basis set size) as AFAIK atomic charges are not a measurable property of atoms in molecules or materials?

Chemists (including myself) like to use them to explain properties like reactivity, but as long as you use them within series most schemes give the similar answers (if you compare trends and changes, not absolute
values)...

Best regards

Peeter

On 28/06/2020 19:56, Thomas Manz thomasamanz(a)gmail.com wrote:
> Hi Partha,
>
> Thank you for your question.
>
> Atom-in-material properties should not explicitly depend on the basis =
> set used to perform the quantum chemistry calculation. The reason for =
> this is that nature corresponds to the limit of an arbitrarily large <= br> > basis set (aka 'complete basis set limit'). For reasons of > computational efficiency, in practical calculations a finite basis set=
> is usually used to get an answer quickly. So, we want to use
> population analysis methods that are not too sensitive to the choice <= br> > of basis set, so that reasonably sized basis sets will produce answers=
> that are not much different than what would be computed if an
> arbitrarily large basis set (aka 'complete basis set limit') w= ere used.
>
> The Mulliken method explicitly depends on the basis set choice and has=
> no complete basis set limit. In other words, the Mulliken populations =
> are not defined for arbitrarily large basis set. This is very bad, > because it means that as the accuracy/precision of the quantum
> chemistry calculation is improved by using larger and larger basis > set, the Mulliken charges get worse and worse with no mathematical > limits or meaning. So, you should not use Mulliken populations. It > also means that if you run a similar calculation twice, but using
> different basis sets for each calculation, the Mulliken charges that <= br> > you get from the two similar calculations may not be similar. For
> example, Mulliken population analysis may predict the cations have > turned into anions, and vice versa, even though it is exactly the same=
> material. In other words, Mulliken population analysis often gets
> cations and anions confused; it does not know the distinction between =
> them.
>
> Between the two methods you mentioned, natural population analysis > (NPA) is the better choice. An even better choice would be to use a > method like DDEC6 for which the atom-in-material properties are a
> functional of the electron and spin density distributions with no
> explicit basis set dependence, and for which the atom-in-material
> properties are chemically meaningful across a broad range of material =
> types.
>
> Sincerely,
>
> Tom
>
>
> On Sun, Jun 28, 2020 at 8:43 AM Partha Sengupta anapspsmo%x%gmail.com
> <= http://gmail.com> <owner-chemistry*_*ccl.net
> <mailto:owner-= chemistry*_*ccl.net>> wrote:
>
>=C2=A0 =C2=A0 =C2=A0Sir, For a metal complexes involving Cu[N,O donor],= [Chlorine and
>=C2=A0 =C2=A0 =C2=A0fluorine atoms are in the benzene ring]. I found th= at cu has
>=C2=A0 =C2=A0 =C2=A00.96091 charges( Natural Population analysis) while= Mulliken
>=C2=A0 =C2=A0 =C2=A0atomic charge is 0.457333. On the same process N ha= s -0.53717 and
>=C2=A0 =C2=A0 =C2=A0-0.29941 respectively. What is the reason behind th= is? What will
>=C2=A0 =C2=A0 =C2=A0be the better choice to represent?
>=C2=A0 =C2=A0 =C2=A0Partha Sengupta
>
>=C2=A0 =C2=A0 =C2=A0--
>=C2=A0 =C2=A0 =C2=A0*/Dr. Partha Sarathi Sengupta
>=C2=A0 =C2=A0 =C2=A0Associate Professor
>=C2=A0 =C2=A0 =C2=A0Vivekananda Mahavidyalaya, Burdwan/*
>

-=3D This is automatically added to each message by the mailing script =3D-=
E-mail to subscribers: CHEMISTRY,ccl.net or use:
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/cgi-bin/ccl/s= end_ccl_message

E-mail to administrators: CHEMISTRY-REQUEST,ccl.net or use
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/cgi-bin/ccl/s= end_ccl_message
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/chemistry/sub_un= sub.shtml

Before posting, check wait time at: http://www.ccl.net

Job: http://www.ccl.net/jobs
Conferences: http://server.ccl.net/chemist= ry/announcements/conferences/

Search Messages: http://www.ccl.net/chemistry/sear= chccl/index.shtml
=C2=A0 =C2=A0 =C2=A0 http://www.ccl.net/spammers.txt

RTFI: http://www.ccl.net/chemistry/aboutccl/ins= tructions/

--000000000000bfdce405a92c442b-- From owner-chemistry@ccl.net Sun Jun 28 20:02:00 2020 From: "Thomas Manz thomasamanz++gmail.com" To: CCL Subject: CCL: Charge Message-Id: <-54120-200628195849-30125-NWn8yzd8lfEzU9sGZmGA3A{}server.ccl.net> X-Original-From: Thomas Manz Content-Type: multipart/alternative; boundary="000000000000b9736a05a92db8ef" Date: Sun, 28 Jun 2020 17:58:30 -0600 MIME-Version: 1.0 Sent to CCL by: Thomas Manz [thomasamanz,+,gmail.com] --000000000000b9736a05a92db8ef Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Hi Robert, >If you see inconsistency across charge schemes, it is not the fault of a given scheme, but the notion itself. I think this all depends on the situation. There may be some situations where two different atomic population analysis methods yield different results for one atom-in-material descriptor (e.g., net atomic charge, atomic spin moment, etc.) and still both values might be considered reasonable as long as each method yields chemically self-consistent results (i.e., that method gave various atom-in-material descriptors that were internally chemically consistent). Yet, we must not forget other situations where the computed result should closely match a chemically well-defined value. For example, in the N^^C60 endohedral complex the electronic state of the N atom is directly observable from spectroscopy to be a quartet state and it is clear from spectroscopic measurements that charge transfer between the N atom and the C60 cage is small or negligible in this material. Therefore, it is clear in this case that any reasonable atomic population analysis method should give small or negligible charge transfer amount between the N atom and C60 cage and should yield atomic spin moments consistent with a quartet spin state of the N atom. It is fair to say atomic population analysis methods that do not do this have given a wrong result. A special aspect of this test system is the N atom is mostly isolated but not fully isolated by virtue of being located in the middle of the C60 cage; this yields a chemically well-defined state for this N atom. One can imagine other well-defined systems that can serve as useful test cases. For these well-defined test cases, it is possible to construct falsifiable scientific tests for atomic population analysis methods. In other words, it is possible in some cases to prove whether an atomic population analysis method yields the right or wrong result. For this reason, atomic population analysis methods can be falsified under some circumstances. Sincerely, Tom On Sun, Jun 28, 2020 at 4:45 PM Robert Molt r.molt.chemical.physics%a% gmail.com wrote: > People can have have discussions of the best charge model as they see fit= . > I make no comment on comparative charge models. > > I do offer the reminder that atomic charge is *not *an observable. This > is a construct in the minds of chemists. It is literally impossible to do > an experiment to measure the notion. One of the many reasons for this is > the fact that one cannot reduce the electron density (a function of 3 > spatial dimensions, at least) to being =E2=80=9Cowned=E2=80=9D by one ato= m, as opposed to > distributed to unequal degrees in space. > > I concur with Dr. Burk=E2=80=99s claim that atomic charges are, at best, = an > internal scheme. If you see inconsistency across charge schemes, it is no= t > the fault of a given scheme, but the notion itself. > > On Jun 28, 2020, at 2:27 PM, Peeter Burk peeter.burk..ut.ee < > owner-chemistry%a%ccl.net > wrote: > > > Sent to CCL by: Peeter Burk [peeter.burk-.-ut.ee] > Hi, > > Can we even discuss which charge partitioning scheme is better (aside of > convergency with increasing basis set size) as AFAIK atomic charges are n= ot > a measurable property of atoms in molecules or materials? > > Chemists (including myself) like to use them to explain properties like > reactivity, but as long as you use them within series most schemes give t= he > similar answers (if you compare trends and changes, not absolute values).= .. > > Best regards > > Peeter > > On 28/06/2020 19:56, Thomas Manz thomasamanz(a)gmail.com wrote: > > Hi Partha, > > Thank you for your question. > > Atom-in-material properties should not explicitly depend on the basis set > used to perform the quantum chemistry calculation. The reason for this is > that nature corresponds to the limit of an arbitrarily large basis set (a= ka > 'complete basis set limit'). For reasons of computational efficiency, in > practical calculations a finite basis set is usually used to get an answe= r > quickly. So, we want to use population analysis methods that are not too > sensitive to the choice of basis set, so that reasonably sized basis sets > will produce answers that are not much different than what would be > computed if an arbitrarily large basis set (aka 'complete basis set limit= ') > were used. > > The Mulliken method explicitly depends on the basis set choice and has no > complete basis set limit. In other words, the Mulliken populations are no= t > defined for arbitrarily large basis set. This is very bad, because it mea= ns > that as the accuracy/precision of the quantum chemistry calculation is > improved by using larger and larger basis set, the Mulliken charges get > worse and worse with no mathematical limits or meaning. So, you should no= t > use Mulliken populations. It also means that if you run a similar > calculation twice, but using different basis sets for each calculation, t= he > Mulliken charges that you get from the two similar calculations may not b= e > similar. For example, Mulliken population analysis may predict the cation= s > have turned into anions, and vice versa, even though it is exactly the sa= me > material. In other words, Mulliken population analysis often gets cations > and anions confused; it does not know the distinction between them. > > Between the two methods you mentioned, natural population analysis (NPA) > is the better choice. An even better choice would be to use a method like > DDEC6 for which the atom-in-material properties are a functional of the > electron and spin density distributions with no explicit basis set > dependence, and for which the atom-in-material properties are chemically > meaningful across a broad range of material types. > > Sincerely, > > Tom > > > On Sun, Jun 28, 2020 at 8:43 AM Partha Sengupta anapspsmo%x%gmail.com < > http://gmail.com> ccl.net>> wrote: > > Sir, For a metal complexes involving Cu[N,O donor], [Chlorine and > fluorine atoms are in the benzene ring]. I found that cu has > 0.96091 charges( Natural Population analysis) while Mulliken > atomic charge is 0.457333. On the same process N has -0.53717 and > -0.29941 respectively. What is the reason behind this? What will > be the better choice to represent? > Partha Sengupta > > -- */Dr. Partha Sarathi Sengupta > Associate Professor > Vivekananda Mahavidyalaya, Burdwan/* > > > > > -=3D This is automatically added to each message by the mailing script = =3D-> the strange characters on the top line to the %a% sign. You can also> > E-mail to subscribers: CHEMISTRY%a%ccl.net or use= :> > E-mail to administrators: CHEMISTRY-REQUEST%a%ccl.net > or useConferences: > http://server.ccl.net/chemistry/announcements/conferences/> > > > --000000000000b9736a05a92db8ef Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable

Hi Robert,

>If you see inconsistency= across charge schemes, it is not the fault of a given scheme, but the noti= on itself.=C2=A0=C2=A0

I think this all depend= s on the situation. There may be some situations where two different atomic= population analysis methods yield different results for one atom-in-materi= al descriptor (e.g., net atomic charge, atomic spin moment, etc.) and still= both values might be considered reasonable as long as each method yields c= hemically self-consistent results (i.e., that method gave various atom-in-m= aterial descriptors that were internally chemically consistent).=C2=A0

Yet, we must not forget other situations where the com= puted result should closely match a chemically well-defined value. For exam= ple, in the N^^C60 endohedral complex the electronic state of the N atom is = directly observable from spectroscopy to be a quartet state and it is clear= from spectroscopic measurements that charge transfer between the N atom an= d the C60 cage is small or=C2=A0 negligible in this material. Therefore, it= is clear in this case that any reasonable atomic population analysis metho= d should give small=C2=A0or negligible charge transfer amount between the N= atom and C60 cage and should yield atomic spin moments consistent with a q= uartet spin state of the N atom. It is fair to say atomic population analys= is methods that do not do this have given a wrong result. A special aspect = of this test system is the N atom is mostly isolated but not fully isolated= by virtue of being located in the middle of the C60 cage; this yields a ch= emically well-defined state for this N atom.

One c= an imagine other well-defined systems that can serve as useful test cases. = For these well-defined test cases, it is possible to construct falsifiable = scientific tests for atomic population analysis methods. In other words, it= is possible in some cases to prove whether an atomic population analysis m= ethod=C2=A0yields the right or wrong result. For this reason, atomic popula= tion analysis methods can be falsified under some circumstances.=C2=A0

Sincerely,

Tom

On Sun, J= un 28, 2020 at 4:45 PM Robert Molt r.molt.chemical.physics%a%gmail.com <owner-chemistry^^ccl.net> wrote:

People ca= n have have discussions of the best charge model as they see fit. I make no= comment on comparative charge models.

I do offer the re= minder that atomic charge is not an observable. This is a construct = in the minds of chemists. It is literally impossible to do an experiment to= measure the notion. One of the many reasons for this is the fact that one = cannot reduce the electron density (a function of 3 spatial dimensions, at = least) to being =E2=80=9Cowned=E2=80=9D by one atom, as opposed to distribu= ted to unequal degrees in space.

I concur with Dr.= Burk=E2=80=99s claim that atomic charges are, at best, an internal scheme.= If you see inconsistency across charge schemes, it is not the fault of a g= iven scheme, but the notion itself.

On Jun 28, 2020, at 2:27 PM, Peeter Burk peeter.burk..ut.ee <owner-chemistry%a%ccl.net> wrote:=

Sent to CCL by: Peeter Burk [peeter.burk-.-ut.ee]
Hi,

Can we even di= scuss which charge partitioning scheme is better (aside of convergency with= increasing basis set size) as AFAIK atomic charges are not a measurable pr= operty of atoms in molecules or materials?

Chemists (including mysel= f) like to use them to explain properties like reactivity, but as long as y= ou use them within series most schemes give the similar answers (if you com= pare trends and changes, not absolute values)...

Best regards
Peeter

On 28/06/2020 19:56, Thomas Manz thomasamanz(a)gmail.com wrote:
Hi Partha,

Thank you for your question.

Atom-in-mat= erial properties should not explicitly depend on the basis set used to perf= orm the quantum chemistry calculation. The reason for this is that nature c= orresponds to the limit of an arbitrarily large basis set (aka 'complet= e basis set limit'). For reasons of computational efficiency, in practi= cal calculations a finite basis set is usually used to get an answer quickl= y. So, we want to use population analysis methods that are not too sensitiv= e to the choice of basis set, so that reasonably sized basis sets will prod= uce answers that are not much different than what would be computed if an a= rbitrarily large basis set (aka 'complete basis set limit') were us= ed.

The Mulliken method explicitly depends on the basis set choice a= nd has no complete basis set limit. In other words, the Mulliken population= s are not defined for arbitrarily large basis set. This is very bad, becaus= e it means that as the accuracy/precision of the quantum chemistry calculat= ion is improved by using larger and larger basis set, the Mulliken charges = get worse and worse with no mathematical limits or meaning. So, you should = not use Mulliken populations. It also means that if you run a similar calcu= lation twice, but using different basis sets for each calculation, the Mull= iken charges that you get from the two similar calculations may not be simi= lar. For example, Mulliken population analysis may predict the cations have= turned into anions, and vice versa, even though it is exactly the same mat= erial. In other words, Mulliken population analysis often gets cations and = anions confused; it does not know the distinction between them.

Betw= een the two methods you mentioned, natural population analysis (NPA) is the= better choice. An even better choice would be to use a method like DDEC6 f= or which the atom-in-material properties are a functional of the electron a= nd spin density distributions with no explicit basis set dependence, and fo= r which the atom-in-material properties are chemically meaningful across a = broad range of material types.

Sincerely,

Tom

On S= un, Jun 28, 2020 at 8:43 AM Partha Sengupta anapspsmo%x%gmail.com <http://gmail.com> <owner-chemistry*_*ccl.net <mailto:owner-chemistry*_*ccl.net>> wrote:

=C2=A0=C2=A0= =C2=A0Sir, For a metal complexes involving Cu[N,O donor], [Chlorine and
= =C2=A0=C2=A0=C2=A0fluorine atoms are in the benzene ring]. I found that cu= has
=C2=A0=C2=A0=C2=A00.96091 charges( Natural Population analysis) wh= ile Mulliken
=C2=A0=C2=A0=C2=A0atomic charge is 0.457333. On the same p= rocess N has -0.53717 and
=C2=A0=C2=A0=C2=A0-0.29941 respectively. What= is the reason behind this? What will
=C2=A0=C2=A0=C2=A0be the better c= hoice to represent?
=C2=A0=C2=A0=C2=A0Partha Sengupta

=C2=A0=C2= =A0=C2=A0-- =C2=A0=C2=A0=C2=A0=C2=A0*/Dr. Partha Sarathi Sengupta
=C2= =A0=C2=A0=C2=A0Associate Professor
=C2=A0=C2=A0=C2=A0Vivekananda Mahavi= dyalaya, Burdwan/*

-=3D This is automatical= ly added to each message by the mailing script =3D-
To recover the email= address of the author of the message, please change
the strange charact= ers on the top line to the %a% sign. You can also
look up the X-Original= -From: line in the mail header.

E-mail to subscribers: CHEMISTRY%a%ccl.net or us= e:
=C2=A0=C2=A0=C2=A0=C2=A0http://www.ccl.net/cgi-bin/ccl/send_ccl_m= essage

E-mail to administrators: CHEMISTRY-REQUEST%a%ccl.net or use<= br> =C2=A0=C2=A0=C2=A0=C2=A0http://www.ccl.net/cgi-bin/ccl/send_ccl_mess= age

http:/= /www.ccl.net/chemistry/sub_unsub.shtml

Before posting, check wai= t time at: http://www.ccl.= net

Job: h= ttp://www.ccl.net/jobs Conferences: http://server.ccl.net= /chemistry/announcements/conferences/

Search Messages: h= ttp://www.ccl.net/chemistry/searchccl/index.shtml

If your mail b= ounces from CCL with 5.7.1 error, check:
=C2=A0=C2=A0=C2=A0=C2=A0http://www.ccl.net/= spammers.txt

RTFI: http://www.ccl.net/chemistry/aboutccl/= instructions/

--000000000000b9736a05a92db8ef--