TOF individual peak fitting showing same profile parameters

cli

Hello all,
I am a novice user of TOPAS and is now trying to fit peak intensity of a set of TOF data.

The aim is to extract peak shape function for individual peak and to investigate the variation with d/tof.
I was following the method from this tutorial for my task

http://topas.dur.ac.uk/topaswiki/doku.php?id=time_of_flight_tof_peak_fitting

The refined profile parameters however are the same for different peaks, although I have put them into different xo_Is and they show up as individual parameters though in the error matrix they do display.

The fitted errors for peak intensity and position also show the same behavior which is very strange…

Can anyone comment on this? Or maybe share with me a .inp file, where individual peak fitting for TOF is successfully?

Code as follows,

r_wp 9.59706967 r_exp 1.51447607 r_p 5.88623739 r_wp_dash 33.9325396 r_p_dash 39.262383 r_exp_dash 5.35476151 weighted_Durbin_Watson 0.12032738 gof 6.33689092
do_errors
iters 100000
no_LIMIT_warnings
chi2_convergence_criteria 0.00001

#define Si_POWGEN_bank1

#ifdef Si_POWGEN_bank1
TOF_XYE (ALL_PG3_37849-1.xye,= Yobs_dx_at(Xo).5;)
TOF_LAM(0.001)
scale_pks = D_spacing^4;
start_X 8000
finish_X 90000
bkg @ 61.3845949`_0.212261848 -0.571076017`_0.385933515 -0.521210336`_0.337763886 -1.41751975`_0.310779123 -0.322338134`_0.249688129 -0.830061073`_0.181707633
neutron_data
local !zero -2.16008 'don't refine
local !dif_c 22592.70827 'don't refine
local !dif_a -8.28523 'refine if necessary
pk_xo = zero + dif_c D_spacing + dif_a D_spacing^2;

str

scale scal_f1 81.9708166`_0.6399
space_group Fd-3m:2
Cubic(5.431178)
site Si num_posns 8 x 0.125 y 0.125 z 0.125 occ Si 1 beq beqSi 0.00388`_0.00830
TOF_PV(si_fwhm, 340.82370`_4.74739, si_fwhm_lor, 1`_0.019429912, dif_c)
'TOF_Exponential(@, 100, ,0, 1, dif_c, -)
'TOF_Exponential(@, 100, @, 100, 4, dif_c, +)

xo_Is xo @ 30659.1137`_6.33689092 update = Val + Rand(-1, 1) Change; I @ 5702.32662`_6.33689 max 30000 TOF_PV(@, 219.79170`_6.33689 max 1000, @, 0.5`_6.33689092, dif_c)
xo_Is xo @ 36972.9316`_6.33689092 update = Val + Rand(-1, 1) Change; I @ 11758.92474`_6.33689 max 30000 TOF_PV(@, 219.79170`_6.33689 max 1000, @, 0.5`_6.33689092, dif_c)
xo_Is xo @ 43350.4812`_6.33689092 update = Val + Rand(-1, 1) Change; I @ 11979.38159`_6.33689 max 30000 TOF_PV(@, 219.79170`_6.33689 max 1000, @, 0.5`_6.33689092, dif_c)
xo_Is xo @ 70760.828`_6.33689092 update = Val + Rand(-1, 1) Change; I @ 4000.64690`_6.33689 max 30000 TOF_PV(@, 219.79170`_6.33689 max 1000, @, 0.5`_6.33689092, dif_c)

Thanks u all for the help!

johnsoevans

I don't think you can mix the peak shape macros in the way you're trying.

I think you need to use something more like the first "trick" on http://topas.dur.ac.uk/topaswiki/doku.php?id=time_of_flight_tof_peak_fitting.

This will let you properly include the instrumental function when you peak fit. This is important with tof data. You're currently refining back-to-back exponentials for this and they seem to have gone to strange values. I suggest you get appropriate values from the instrument scientists and fix them (at least initially).

You should then be able to use something like the code sections below. You might want to change the macro so it allows different fwhm's for each peak. At the moment it uses a simple size/strain type parameterisation across the whole pattern.

John

**********************************************************************************
r_wp 4.54083053 r_exp 1.5140991 r_p 2.92185396 r_wp_dash 16.517065 r_p_dash 21.8919292 r_exp_dash 5.50746678 weighted_Durbin_Watson 0.2848859 gof 2.99903125
do_errors
iters 100000
no_LIMIT_warnings
chi2_convergence_criteria 0.0000001

#define Si_POWGEN_bank1

#ifdef Si_POWGEN_bank1
TOF_XYE (ALL_PG3_37849-1.xye,= Yobs_dx_at(Xo).5;)
TOF_LAM(0.001)
scale_pks = D_spacing^4;
start_X 8000
finish_X 90000
bkg @ 62.0886548`_0.112162635 -2.13135162`_0.200687157 1.14665218`_0.181433041 -0.786478208`_0.170901176 0.742346841`_0.161535983 -0.544177666`_0.151754591 0.173579161`_0.148072372 -0.0444392671`_0.126316794 -0.370325647`_0.088300723
neutron_data
TOF_x_axis_calibration(zero, -10.27386`_0.37826, dif_c, 22627.07441`_6.66155, dif_a, -46.38715`_2.41459)
TOF_Exponential(@, 0.00000`_186741.89747,@ , 16102.41566`_524099.14403, 1, dif_c, -)
TOF_Exponential(@, 4.02588`_0.09069, @, 284.92238`_3.47087, 4, dif_c, +)

str
scale scal_f1 80.9201816`_0.4155
space_group Fd-3m:2
Cubic(5.431178)
site Si num_posns 8 x 0.125 y 0.125 z 0.125 occ Si 1 beq beqSi 0.17466`_0.00501
TOF_PV(si_fwhm, 194.73217`_16.18474, si_fwhm_lor, 1`_0.0611965913, dif_c)

'peaks here

prm dsp 0.00000`_24.79753 min 0 max 500
prm dspsq 54.87860`_11.23441 min 0.1
prm lor 0.19460`_0.13129 min 0 max 1
tof_peak_fit2(peak1_dsp, 3.10855`_0.00061, @, 756.60686`_36.86218)
tof_peak_fit2(peak2_dsp, 1.91194`_0.00024, @, 1639.01367`_82.59958)
tof_peak_fit2(peak3_dsp, 1.73173`_0.02188, @, 7.44876`_27.07633)
#endif

/* Modified peak macro without dspacing in hkl_Is phase to fix minor bug */
macro tof_peak_fit2(p1,v1, p2,v2) {
#m_argu p1
#m_argu p2
If_Prm_Eqn_Rpt(p1, v1, min =0.9*v1; max =1.1*v1;)
If_Prm_Eqn_Rpt(p2, v2, min .000000000001 max 1e9)
hkl_Is phase_name CeV(p1,v1) a =10*CeV(p1,v1); b 0.1 c 0.1 space_group "P1" load hkl_m_d_th2 I { 10 0 0 1 0 80114.92 =CeV(p2,v2); }
peak_type pv pv_lor =lor; pv_fwhm = dsp*CeV(p1,v1) + dspsq*CeV(p1,v1)^2;
}

cli

Dear John,

It works out beautifully, and is exactly what I am looking for!

Some notes for those who are interested in using this method to carry out individual peak fitting,

- the p1 parameter in macro tof_peak_fit2 macro is in d spacing rather than TOF

- it's always good idea to restrict the range of your parameters, one can always find d from hkl_Is, then give it like 2% error so that the position of the peak is fixed.

ChengLi

johnsoevans

Sorry for the silly parameter names - I posted from Schipol airport after a 13 hour overnight flight. I modified the labels to say "peak1_dsp" in the original post. You could rewrite the macro to work in tof, but d-spacing is normally more convenient for e.g. indexing.