Volun1c
108. 11111nbcr 4
Clll’hflCAL
I’IIYSICS
LETTERS
HIGH-RESOLUTIONNEUTRONINELASTICSCATTERING
13 July
1984
FROXIHEXAMETHYLENETETRAMINE
H.J. LAUTER
latcd
1. Introduction
rcccnt
tltcorctical
work
011
the vibrat ions1 intcn-
sitics [9--l I]. Neutron inelastic sxattcring(NlS) from hydrogcncous materials is ;I sensitive nictliod for studying the vibrational tnodcs of tnolccular systems. There arc scvcral feattms which make NIS com~~lctncntar~ to the infrared and Ramt~ tcchniqucs, e.g. the lack of optical selection rules and the possibility to co111putc both the frcqttcnaies and the NIS itttcttsitics of tltc normnl modes. I-Iowcver. tlw niain dis3dvmt3gc of NIS is rite relatively poor rrsoltttion which has been available at high cncrgy transfers.
3. Esperinlental
The principle of tllc hcryllitm filtrr detector spwtronwtcr has been described [ 131. The berylliunl filtc~-. sitwtcd bctwccn the smiplc and the dctcctor. allows
neutrons whose energy (after scatis smillur than tlic cur-off cncrgy of E, = 42 unt to bc tratmnittcd. Provided
at the hot source (IN I ) at
(ILL). Dedicated ~nonoclirotiiators for the (200). (220) and (331) reflections of Cu combined with the lower cur-off energy of a graphite filter, allowrd 11s to obtain tsccllcnt resolution in the region 250-2000 cni-t , In order to demonstrate the improvcmcnt of the resolution, WC have reinvestigated the NIS spcctrun~ of hcsa!~letllylcnetetr~!~~i~~e (HMT). WC 1mw cl~oscn this co~npound because it has Arcady been well studied by optical spectroxopies [l-6] and by NlS [7,S]. and also because the NIS spectrum of HMT has stimu-
0 009-2614/84/S
03.00 0 Elsevier Science Publishers
those
that the other contributions to the overall resolution (monocliroliiator. collitiiations) are smaller. tilt slxctral resolution cm bc improved by using a filter with a lower cut off. (For ;1 cAxthtion of a filter imtru-
We report in this paper a major improvcnient oi the spectral resolution that has been obtained using
IIIC IWW monocllromator the Institut Law-Lmqpin
only
tcring with the sample)
nwnt, xc ref. [ 131 - in particular for the old Befilter version at tlic ILL.) TIIC different properties of various tnatcrials for tltc design of cold neutron iiltcrs have been discussed by Egelstaffand Pease [ 141 and more rcccntly by Tho~nns [ IS]_ A graphite filter is x1 intcrcsting possibility since this material has a low cut-off energy of HOWI:;- = 14 cm- ’ given by the (001) reflection. ever, tlic transmission for graphite isabout half that of Be and the leakage (trammission above k?c) is higher for grap!iitc [ 151.
B-V.
volulne
108. llulllbcr 4
A graphite filter was made at Hanvell scvcral years ago. The Inaterial used for the polycrystalline filter was reactor-grade graphite and cadmium sheets were placed
within
the filter
to reduce
the multiple
the signal-to-noise ratio by a factor 3 with respect to only the graphite filter. The transmission of the graphite I‘ilter has been nleasured in adding it to the Bc filter. It amoutlts only to 7% (for the signal). which is partially due to the fact that the cross-scctional arca of this filter is snialler than that of tlic Bc filter. The background was lowered by a f‘aclor 7.5. The scattering angle was always 90”.
scat-
tering. Preliminary tests indicated that a better rcsolutiou could bc obtained at a cost of a factor ==I0 in the peak signal intensity [ 161, For this reason, the g~apliilc filter w3s not used any more at I-Iarwcll. The lowcl tiansniission of graphite compared to Bc is 1101 :I problcnl on the new IN1 B spectrometer since tlw irlcidcnt flux is very large. This was made
3. Results
pmsibic by vertical t‘ocusing of the nwnocl~romator crystals amI also bccausc of the better rcllcctivity of 11~ crystals which arc now used iii reflection md not iii traiislnissioii as in llic old version. The situation was 1l1us IWIC favorable to test :I graphite I’iltcr (the
I larwll
spectra of HhlT at 5 K arc SIIOW in figs. (tlic tolal n111 time was 9 II). From 750 to 900 cm- I, tlic data wcrc obtained with llic (270) IIlanc ol‘ tllc copper monochro~llator. TIIC resolution of this iiionocliroinnlor is worse than that of tlic sccondary spcclroiuclcr at liiglicr cncrgics. so that the The NIS
I
filler was Icut to us by Mr. D.1i.C. Ilarris).
To ovcrconw tllc problem ol’lcakagc ol’ tlw giapliitc filter. WC have pluccd in I’rout of it a Bc filrur (boll1 fillers wcrc cooled to liquid nilrogc:cli tcnipcr:lturc). WC I’c~iincl lliat this wnil~ilialioli incrcxcil
300 3500- $ : s _,L 2 z 25oo-
1500-
13 July 1984
Cl-lE%llCALPHYSICS LETTERS
I
and
7
use of a higlicr-order plan12 - tlic Cu (33 1) - yields a bcttcr icsolutiw abovc 5.50 ciiil, with a rcduclioii of the incident Ilux by :I factor 3. Altllougl~ we have
500
I
700
I
!300 Etcm.’
U‘
r
1500
1000
Volun~c
CHEhflCALPHYSICS
108. number 4
0
,
I 30 I:@. 2. NIS spectrum
13 July 1984
LETTERS
1 50
I 40
of IlhlT 31 5 K obUincd
1 60
EUHrl
with the CU (331) plans.
recorded the spectrum up to 4000 CI~-~, with the Cu (331) plane, only the range 800-2200 c~n-~ is
These resolutions are of the mm order as the optical ones. This unexpected high resolution might be due
shown in fig. 2 because the resolution is only slightly improved above 7200 cm- * compared to ref. [8].
to the higher absorption
The use of a collimator bctwccn the monochronlator and the sauiple would further iniprove the resolution for higher frequcncics. Thr, background obscrvcd at high energies is due to the niultipi~onons and to fast neutrons due to a not yet optimized protection. It cm bc noted that the contamination from sccondorder neutrons is negligible since we do not observe around 350 cm- 1 fcaturcs corresponding to the intcnsc peaks situated front 1300 to 1500 CII~-~ but more accurate tests will bc done.
4. Discussion The comparison with the NIS spcctrunl of ref. 181 obtained at the sync tcmpcraturc of 5 K shows that the spectral resolution has been n~uch improved (the number of observed pc:ks hs more than doubled). Tltc maarrcd resolution AK/E, is 3!%with the Cu (230) plme uild 2% with tlw Cu (331) plant (up to I500 m- *). Thcsc spectra have the best NIS rcsolution ruportcd so far. The full width at half maxinlunl (fwlini) of tlic peak at 380 cm- * (vlG) is only 11.5 cm - ’ ) ilnd 20 cn1- 1 for the pwk (11930 cm- ’ (q 5).
for long wavelengh
neutrons
in the two filters. Our observed fwhm and the calculation of the resolution of the monochronlator show that the intrinsic widths of the intramolecular modes are very mall. This not only indicates very little dispersion of these modes but also that WCare observing the fundamentals and not the broadened components [ 1 11. For a sniallcr molecular niass, tlic scattering is quite different [ 171. The improved spectral resolution makes the nssignment of the NIS spectrum easier (it is indicated in figs. 1 and 2). To support our assignment, we hove calculated the NIS intensities from the force fields given in refs. [5,7]. In view of a few disagreements with the assignments used for these normal coorditlutc analyses (rminly for the inactive Ft modes), WC intend to take up this calculation again. An intcnsc peak corresponding to the vt 5(Fi) mode is found at 930 cm- 1 ; although our precision is only of=1 5 cm- 1 instead of =I cm- 1 as in infrared or I~uman. WCthink that IJ, s has to be placed higher thm 9 17 cm- t [G] (it is ass&nod at 925 ~111”1 in rcfs. [3,5]). The frcqucncy diffcrcncc isgreatcr for yt4(F1): nothing is observed in the NIS spectrum at 11SO cm- 1 [S,6] but an iutcnsc peak is obtoincd at 395
Voln~nc I O&t.nu~nbcr
(‘III:hllCAL I’IIYSICS Llil‘l’lil~S
4
1060 cni - I wliiclt cm bc readily wigtIed to IjI J (it is lhwd at 1076: CI~- 1 in ref. [3] j. The I), SW,) nwclc is lh~cd 31 1306 cni-- 1 111 rc1'. [O] and at 1315 cm- t in ref. [SJ: WCfind that ;I value or*!340 c111- ' would tilt
this
i 270
~111-
1 f0r
WC conl’irm
~11~ I+,)
1110d~
[s,h]
;
is ckirly scp:tratcd fronl Y, 1 it1 fig. 7. We wit11 a wluc of 1301 cn- 1 for rjl 2 [5,6] bur
niodc
agree 011r
witlt tltc NIS s!xct~t~~~I.
fit bcttcr
V~I~UC d
prcliniitt3ry
calculations
I for u^, [56]
1459 cmis situated Several
indic:ttc my
have
at 1428 all- I in ref. [;I OVCI-tones and con~bination
11~11tlic value of lo bc
cllangcd
I3 July 1984
IliIl3tioIl. 011 a reactor, yields a vety good resolution, even at the high A’ values involved. This resolution is now siil‘ficient to follow the relative weights of the funda~nc~lt;~i peaks and of the side peaks as a fu~lc-
tioti Of tfZlll]Xl’iItlIrC. For large ~nolcculcs,
the profile of thr NIS spcctra. obtained at wry low tcmpcrntttrc, ~311 be favorably cotllpitrcd to the CiI!CtII;ItCd OIlC-]lhOIlOIl S]lCCII 11111.
(il
j.
peaks
cali
be
in the NIS spcctrun~. In particular. 111any co~llhialions with he latlicc modes arc 110~ well rcsolved and ir is k11own Illat the density of states of tltc lattice ~nodcs ilus two maxim at wL = 40 ant1 60 CIlt - 1 17, IS] _ These combinations arc indicated by arrows in figs. 1 arid 7. Tllc intensities of the side lx&, upproximately A40 and 560 cm- 1 nwtly from
Acknowledgement
four~d
Ill? flIIltlaIneIltal
I’t6,
are governed
ol- vt 6.
5. Conclusion It is clear that in NIS. high resolution is needed to separate the different vibrational modes. We have found that for HMT, a clearer assignnwnt could be obtained and this should lead to a better evaluation of the force field. It also appears that the Be + graphite filters COIII-
396
References
by the popiIlatioIi
factor. For the other fundantcntals at frequencies uh (e.g. a~ 506, 678, S20 and 1060 cull- ') only the combinations wA + wL can bc observed. The intensity o1‘tl1csc conlbinations is proporIiona1 to K4, whcrc El is the ncutron n~omentunl transfer. Tiwrefore. thy are more intense at large energy (nio~iicnttun) transfers. However, the intensity of the fundamentals. even if multiplied by the Debye-Wailer fattar, is always more intense than the side peaks. This is even rile case for the modes situated above 1000 ull* where the ET values are very large (K = 9 :I- 1 31 w = 1400 ci11- I)Other combinations between intense NIS peaks cm1 be assigned. e.g. v 16 + vz5 at S95 cni- I, dcf.(CH?) + vt6 at 1860 c111-~ and def.(CH,) + uq5 at 1950 c111- 1, where def.(C:-I,) represents the conipositc band centered at ~1450 CM- l _Finally, the weak band observed at 760 cni corresponds to the first OVcI’tOIle
We are very 1111Icl1 indebted to Mr. D.H.C. Hnrris for rhc loan of the graphite filter. WC also thank Dr. RI. Gliosli for iiclpfttl discussions.
] I] L. Couture-Xlatllicu, J.1’. Itlathicu. J. Cremer and II. I’outeI. J. Chim. I’llys. 48 (195 1) I. [ 31 A. Chcutin and J.P. blnrhicu, J. Chim. Ply. 53 (1956) 106. ]3] R. Mcckc and Ii. Spicscckc. Chcm. Bcr. 88 (1955) 1997. Spcctrochim. Acta 7 (19.56) 387. ]4] 1’. Shire. S. Nnkamura. II. Xlumtu and II. Ncyitu. J. sci. iliroshima Univ. A II 31 (1967) 131.
] 51 J.1:. tkrtic
and M _Solinns. J _ Chem. Phys. 6 1 (1974) 1666. ]6] 1I.P. ChCdin, J.P. tc Rolkmd, C. Capderroque and G.G. Dumas. Mol. Cryst. Liquid Crysr. 78 (1981) 67. ]7] M.W. Thomas and R.E. Ghosh. Mol. Phys. 29 (I 975) 1489. [ t;J II. Jobic, R.E. Ghosh and A. Rcnouprez, J. Chem. l’hys.
75 (I 981) 4025. [ 91 A. G&Tin and J. Jobic, J. Chcm. Phys. 75 (1981) 5910. [IO] B. Dorner and A. Griftim. J. Chcm. Phys. 78 (1983) 890. [ I l] 11. Warner, S.W. Lovescy and J. Smith, Z. Physik B51
(1983) 109.
[ 171 B_T.hI. Wiltis. cd., Chemical
applications of thermal neutron scattering (Oxford Univ. Press, London, 1973). ] 131 R. Almairac, J-L. I’refaur, XI. Galtier, C. Bcnoit, A.
hlontaner and H.J. Lauter, hlol. Cryst. Liquid Cryst. 69 (1981) 177.
[II] [ 1.51
[ 161 [ 17) [ 181
P.A. 13gelstnff and R.S. Pease, J. Sci. Inst. 31 (1954) 207. h1.W. Thomas. AERE liarwell Report, lINBFC/P57 (1976). C. Dufil, unpublished results. II. Jobic,Chcm. Phys. Letters 106 (1984) 321. G. Dolling and B-51. Powetl. Proc. Roy. Sot. A3 19 (1970) 209.