how does VASP decide unit cell structure?

[fzcsydl]

When I do interface simulation using supercell, I found that the supercell structure in the POSCAR file is different from the supercell structure in CONTCAR. This means VASP will choose the supercell structure based on you input POSCAR.
My question is how VASP choose unit cell structure? So I can get same unit cell structure in CONTCAR as in POSCAR.

Thanks a lot.

[alex]

I've never seen that behaviour before. Could you please post POSCAR and CONTCAR (the header should be sufficient).
Cheers,

alex

[fzcsydl]

[quote="alex"]I've never seen that behaviour before. Could you please post POSCAR and CONTCAR (the header should be sufficient).
Cheers,

alex[/quote]

I guess the reason may be I choose the unit cell structure is not within the volume of the unit cell I defined. But I am not sure. Hope you can help me.

POSCAR:
1.000
2.676743 -2.666614 -0.000425
2.690507 2.678996 -0.000810
-0.009905 -0.006492 48.466054
27 6 2
Cartesian
13.56281668 -0.04674612 2.77532941
13.55892822 -0.04842193 7.96803872
16.25013277 2.62997871 5.36416351
13.56194234 -0.04357730 13.29157686
16.25419865 2.63403966 10.60339518
10.87939991 -0.05254641 5.33874706
13.56414209 2.63178203 2.68072232
10.87730183 -0.05162880 10.57771530
13.56853690 2.62429391 7.97237667
10.86915841 -0.04371767 15.60865515
13.56930496 2.62709212 13.15251834
13.52804137 -0.01351584 0.33787698
14.91305258 1.28642395 -1.01661559
10.88404702 -0.05888522 -2.37103239
12.22248683 1.28699807 -3.73268421
13.56073820 -0.04614822 -5.09795922
14.90599804 1.29273489 -6.45731163
10.87765414 -0.05201714 -7.82303820
12.22203871 1.28779790 -9.18537884
13.56062397 -0.04510722 -10.55038272
14.90758629 1.29255083 -11.90743093
10.87917220 -0.05231258 -13.26989890
12.22666873 1.28504282 -14.63013606
13.56560827 -0.04680594 -15.99618174
14.90654225 1.29728883 -17.35346635
10.87648299 -0.04696884 -18.71218997
12.23573622 1.27806908 -20.05305391
14.89788137 1.30018364 1.37308653
14.90397050 1.28912484 6.65747934
14.90887019 1.29554593 11.88428162
12.22532590 1.28771459 4.04009752
12.22169907 1.28599633 9.28312508
12.22028550 1.28830870 14.55251413
13.00393563 0.36901899 -20.96054221
11.49876121 2.15585544 -21.01797696
CONTCAR:
1.00000000000000
2.6767430000000001 -2.6666140000000000 -0.0004250000000000
2.6905070000000002 2.6789960000000002 -0.0008100000000000
-0.0099050000000000 -0.0064920000000000 48.4660539999999997
#####
27 6 2
Direct
0.5329817923874194 0.5139058402831529 0.0577202097772072
0.5454299181139921 0.5153266252719658 0.1676914923368658
0.5327520504233535 0.5147614800923596 0.1125738644526341
0.5539078813886094 0.5150648338056116 0.2799323677628741
0.5456003478477884 0.5145989763889312 0.2231582816866348
0.0411434763957443 0.0143093547034890 0.1119383210072871
0.0291087573921873 0.0138423904328613 0.0568862981718668
0.0518217246712896 0.0152437515756994 0.2223235235312324
0.0409075703196919 0.0150884256551207 0.1679889707216216
0.0552781532171628 0.0161789372285314 0.3275626164950525
0.0500686040316083 0.0154996809796084 0.2765349185773281
0.5262863281981747 0.5142775303693072 0.0068130216609465
0.5129293945209971 0.0137816060253740 0.9784122104678401
0.0122272390638786 0.0118462295465066 0.9496966669812053
0.0115521304944500 0.5114932857162580 0.9211793835577557
0.5113242044033539 0.5111534596251165 0.8926102066497147
0.5116765007923799 0.0109187082393531 0.8641997146540951
0.0116983500036638 0.0105232494697644 0.8356691820228187
0.0138946550250444 0.5102838528376004 0.8072927666112319
0.5140053911905018 0.5098011400043113 0.7787586928839072
0.5145874566540435 0.0095195950645494 0.7503716467668758
0.0143666391870208 0.0088917475625159 0.7218685175000848
0.0139507138832995 0.5086334485195125 0.6935464551717828
0.5132327986188507 0.5079360676341156 0.6650452251429774
0.4971798625613932 0.0077846632468723 0.6366975119611573
-0.0045935629912051 0.0072949255952088 0.6082196467226616
0.0952405005948872 0.5068707094614671 0.5817025405063041
0.5276322048097408 0.0140103931582533 0.0288794758755039
0.5413337610425878 0.0151361752750297 0.1401525126692341
0.5506602524267128 0.0146849842684012 0.2498568589263791
0.0330083960213547 0.5142913567077894 0.0847233837993478
0.0455038556690111 0.5152167646019438 0.1955741946945285
0.0545777114338975 0.5160442841621897 0.3058879421546704
0.4906648133904205 0.5069210041902303 0.5787540852616970
0.9590901574763873 0.5068077046190506 0.5526561406120752

[alex]

Looks reasonable. VASP transfers your cartesian input into direct (fractional) coordinates in the first unit cell. Try to visualize CONTCAR and check if it still the same (p4vasp, molden etc do the job).

Cheers,

alex

[forsdan]

@fzcsydl - I would advice against specifying your coordinates the way you do it in the POSCAR file. Your x-coordinates are several times the basis vectors outside your supercell. They will of course be translated back to the specified cell, but it's hard to determine from your entries if atoms will overlap. In addition, comparison to the CONTCAR (where every coordinate is translated back to the cell) can be cumbersome.

So my advice is to specify all the coordinates inside the specific cell from the start.

Cheers,
/Dan

<span class='smallblacktext'>[ Edited Thu Jun 17 2010, 10:21AM ]</span>