Raman Spectroscopy Using a Gas Laser J.P

Raman spectroscopy using a gas laser J.P. Russell

To cite this version:

J.P. Russell. Raman spectroscopy using a gas laser. Journal de Physique, 1965, 26 (11), pp.620-626. 10.1051/jphys:019650026011062001. jpa-00206048

HAL Id: jpa-00206048 https://hal.archives-ouvertes.fr/jpa-00206048 Submitted on 1 Jan 1965

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. 620 dotted line the intensity ratio here is about 3.2 Discussion times too large. The intensity anomaly is even greater in the case of deuterium where in normal M. THEIMER. - Is the interpretation of the deuterium the observed ratio is as great as 9 times Raman spectrum of hydrogen in terms of exci- the value expected on the basis of isotropic scat- tons different from that in terms of lattice vibra- tering only. Up to the present no plausible expla- tions ? It appears that the assignment of the nation of this curious effect has been forthcoming. different peaks to different nearest neighbor ratios It might be noted that the anomaly is not present of ortho and para molecules is not quite consistent in the high pressure gas at 300 OK and 85 °K. with the concept of lattice vibrations.

BIBLIOGRAPHY

[1] CLOUTER (M.) and GUSH (H. P.), Phys. Rev. Letters [5] Mc KAGUE (A. H.), M. A. Thesis, University of Toronto (in press). 1965. [2] GUSH (H. P.), HARE (W. F. J.), ALLIN (E. J.) and [6] VAN KRANENDONK (J.), Physica, 1959, 25, 1080-1094. WELSH (H. L.), Can. J. Phys., 1960, 38, 176-193. [7] VAN KRANENDONK (J.), Can. J. Phys., 1960, 38, 240- [3] BHATNAGAR (S. S.), ALLIN (E. J.) and WELSH (H. L.), 261. Can. J. Phys., 1962, 40, 9-23. [8] STOICHEFF (B. P.), Can. J. Phys., 1957, 35, 730-741. [4] SOOTS (V.), Ph. D. Thesis, University of Toronto, 1963. [9] MAY (A. D.), VARGHESE (G.), STRYLAND (J. C.) and WELSH (H. L.), Can. J. Phys., 1964, 42,1058-1069.

LE JOURNAL DE PHYSIQUE TOME 26, NOVEMBRE 1965,

RAMAN SPECTROSCOPY USING A GAS LASER (1) By J. P. RUSSELL, The Royal Radar Establishment, Great Malvern, Worcs, England.

Résumé. 2014 Le laser à hélium-néon présente, sur la lampe à vapeur de mercure généralement utilisée, l’avantage de donner un rayonnement plus intense, de longueur d’onde (6 328 Å) plus grande, à laquelle les milieux solides étudiés sont plus transparents. En outre, ce rayonnement est spontanément polarisé. Le montage expérimental indiqué, formé d’éléments existant dans le com- merce, permet d’observer la radiation diffusée à 90° et, légèrement modifié, la lumière diffusée vers l’arrière par le silicium, sous un angle de 45° environ. On a ainsi obtenu les spectres Raman du second ordre de divers matériaux solides, dont le tungstate de calcium, le fluorure de calcium, le phosphure de gallium et même, pour ce dernier, le spectre Raman à trois phonons. Deux tableaux résument les résultats expérimentaux. Abstract. - The helium-neon gas laser has many advantages over the mercury arc, previously used. It is more intense, its light has a longer wavelength (6 328 Å) and is plane polarised. The experimental arrangement is described : it is made of commercially available apparatus. The scattered radiation is normally observed at right angle to the direction of the laser beam, but little modification allow the observation of backward scattered radiation, the angle between the laser beam and the direction of observation beeing about 45°. Using this arrangement, the first and second order Raman spectra of several materials, as calcium tungstate, calcium fluoride and gallium phosphide and even, for GaP, three phonons Raman scattering, has been measured. Two numerical tables summarize the experimental results.

1. Introduction. - Since Raman scattering was taline solids. In a solid the scattering is due to first reported in 1928 (C. V. Raman, 1928) measu- the interaction of the light with the lattice vibra- rements of Raman spectra have been a valuable tions or phonons. The first order spectrum which tool in the investigation of vibrational and rota- is due to the creation or destruction of single tional energy levels of molecules and the deter- optic phonons at the center of the Brillouin zone mination of lattice vibration frequencies of crys- identifies the zone center phonons unambiguously. The second order is due to (1) British Crown Copyright. Reproduced with the permis- spectrum processes sion of the Controller, Her Britannic Majesty’s Stationery involving pairs of phonons of equal or equal and Office. opposite wave vector. It provides information

Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:019650026011062001 621 which is complementary to that obtained from two of the disadvantages of the old technique. It phonon infrared absorption and is directly related operates at a longer wavelength, it is intense, plane to the two phonon density of states. The comple- polarized and makes low temperature measu- mentary nature of the two measurements arises rements easy. from differences in selection rules for the two pro- In developing our technique for Raman spec- cesses. (This is particularly true in a lattice like troscopy we have made measurements on a number CaF2 which has inversion symmetry, then because of materials. We will illustrate the advantages of of parity conservation the Raman process can the laser by reference to each of these materials observe even parity phonons and phonon combi- and then describe some new results which have nations while the infrared absorption involves only just been obtained on GaP at 20 OK. odd parity phonons. In this case both sets of measurement are necessary for a complete picture 2. Experimental. - The equipment which we of the phonon spectrum.) have used is all commercially available. The laser Raman scattering probabilities are low, typical is a model 116 Spectra Physics helium neon gas values being 1 : 106 for first order scattering and laser, which we operate at 6 328 A where it gives 1 : 109 for second order scattering. This means a uniphase output of 25 to 30 milliwatts. The that to observe Raman scattering very intense spectra are measured using a Hilger and Watts sources of monochromatic radiation are required. Raman spectrograph with an aperture of approxi- It is usual to use one of the intense emission lines mately F /6 and dispersion of 40 A/mm at 6 328 A. obtained from a gas discharge as the exciting radia- The experimental arrangement is shown in tion. Generally a mercury discharge lamp is used, figure 1. The parallel beam output from the laser this has intense lines at 2 536 A, 4 047 Á, and is passed through a series of baffles to eliminate 4 358 as. The discharge lamp has proved useful as the fluorescent radiation from the collimated laser a source of exciting radiation, however it suffers beam. The small amount of fluorescence which from a number of disadvantages which have tended remains is used to calibrate the plates. The beam to restrict the full development of Raman spec- is passed vertically through the sample, it is focus- troscopy. sed by a short focal length lens so that inside the The laser has provided us with a new source of crystal the mean diameter is less than 100 microns. exciting radiation. It has many advantages over For measurements at room temperature the crystal the mercury lamp and in fact suffers from none rests on a mirror which reflects the laser beam

FIG. 1. 622 back along the same path. At low temperatures Rayleigh scattered radiation (radiation which one face of the crystal is silvered and attached to a has been scattered without frequency change) can cool finger. This arrangement again reflects the be very troublesome in measuring Raman spectra. beam along the same path. In both cases an Using the laser there is not equivalent filter to the external resonator is formed with the output self absorption of Hg 2 536 by mercury vapour. mirror of the laser. This is not a true optical We have found that by placing a small absorbing resonator, however we estimate that as many as shield in the plate holder at the position corres- 5 passes occur. Allowing for the very lossy nature ponding to 6 328 auk most of the Rayleigh scattered of the system this effectively increases the useful radiation can be removed. This technique while beam intensity by a factor of two. effective does prevent small frequency shifts being The scattered radiation is normally observed measured. For shifts greater than 200 cm-1 a at right angles to the direction of the laser beam, multi layer dielectric filter can be used to remove see figure 2. However, in the case of silicon which Rayleigh scattered radiation.

3. Comparison between the laser and mercury discharge. - 3.1. WAVELENGTH. - The laser operates at 6 328 A, which is considerably longer wavelength than the normal mercury lines. This extends the range of materials which can be inves- tigated as many materials are opaque at the shorter wavelengths. The best example is Gallium Phos- phide which is a III/V semiconductor with a band gap of 2.3 e. v., it is opaque at the mercury wavelengths. Using the laser we have measured both the first and second order Raman spectra (M. V. Hobden and J. P. Russell, 1964). It is very strong Raman scatterer, its first order spectrum FIG. 2. being more than 10 times as strong as the first order spectrum of diamond. In fact we have was recently observed 3 phonon Raman in is opaque at16 328 A a different technique scattering The from the GaP, which we believe is the first time 3 phonons employed ; see figure 3. output scattering has been observed. As we are going to report new measurements on GaP at the conclusion of this paper no further discussion will be given now. Raman scattering probabilities decrease as 1 /X4, this could be a serious disadvantage in the use of the laser. However, we have found that the high intensity more than overcomes any decrease in scattering probability.

3.2. INTENSITY. - Undoubtedly there is much more power available in the emission lines from a 3 or 5 kilowatt mercury lamp than the 30 milliwatts we have available from the laser. However, as we use the 30 milliwatts efficiently it appears that there is more useful power available from the laser. This is clearly demonstrated by our measurement FIG. 3. of the second order spectrum of CaF2. The first order spectrum of CaF2 consists of a single line. laser was focussed on the surface of the silicon and This identifies the degenerate fi02 and L02 the radiation back scattered from the surface branches at the center of the zone, the frequency analysed with the spectrograph.’ In this case the shift of 322 cm-1 is very well known. However, angle between the laser beam and the direction of nobody has reported the second order spectrum observations was about 450. until our measurement with the laser (Russell, Recording is done photographically using Kodak 1965). type 103 aF plates which are designed for long integration times and have the best sensitivity at 3.3. POLARIZATION. - The output from the laser 6 328 A. is plane polarized, while the mercury lamp output 623 is unpolarized. The polarization together with and there are 24 distinct branches at the center negligible beam divergence of the laser makes it of the zone. Barker (1964) has shown that the very easy to measure the polarization of the Raman symmetries are Au + Eu for the long wavelength scattered radiation for different crystal orienta- acoustic phonons and tions and of radiation. From polarization exciting 3A. + 5Bg + 5Eg + 4Au + 3B. + 4Eu such a measurement the group theoretical symmetries of the Raman active phonons may be deduced. for the long wavelength optic phonons. The E It is of course possible to polarize the mercury symmetries are 2 fold degenerate, the A and B lamp output and make a similar measurement, symmetries not degenerate. The even parity pho- however with the laser it is much easier. nons (subscript g) are Raman active, the Au are This technique is well illustrated by our recent infrared active for the extraordinary ray, the Eu measurement on calcium tungstate (Russell and for the ordinary ray. Loudon, 1965). There are two Ga’W04 units in the We have detected 11 of the 13 possible Raman primitive cell ; this could lead to 33 optic phonon active phonons and measured the symmetries of 10 branches. However some degeneracies occur of them. Table 1 shows the phonon energies and

TABLE I

symmetries deduced from our Raman measure- spectra of a number of opaque materials, including ment and infrared measurements (Barker, 1964). even metals, may be measured in this way. It is seen that these divide into two groups having frequencies in the ranges 0 - 435 cm7l and 3.5. Low TEMPERATURES. - Using the laser it 778 - 912 cm-l. We believe that the high fre- is comparatively easy to measure Raman spectra at quency set of vibrations correspond to internal low temperatures. Potentially this is very impor- vibrations of the tightly bound W04 ions. The tant as the analysis of a second order Raman low frequency set may include some W04 ions spectrum is easier if there are no difference bands. vibrations, but it is likely that they are mixtures Also there are interesting possibilities in Raman of internal and external vibrations so that no clear scattering due to electronic transitions and spin physical interpretation is possible. waves (Elliott and Loudon, 1963) which require 3.4. Focus. - The laser beam may be focussed low temperatures to be observed. Low tempe- to a very small spot ; at this spot the power density rature measurements have been made using dis- is very high; even using poor optics a power density charge lamps; however the large amount of back- of a least 103 watts jsq . cm is easily obtained. A ground fluorescence makes them quite difficult to similar power density is very difficult to obtain do. - using a mercury lamp. We have made use of Section 4 contains a description of the Raman this property to observe Raman scattering in sili- spectrum of gallium phosphide at 20 OK ; this will con (Russell, 1965). Silicon is opaque at 6 328 A serve as an illustration of the power of low tempe- with an absorption coefficient oc = 2 X 103 ern7l. rature measurementg. However the power density at the focus was high enough that we were able to detect Raman scat- 4. The Raman. spectrum of gallium phosphide tering in the back scattered radiation. The value at 20 OK. - In analysing the second order spectrum obtained for the first order Raman shift of of GaP at room temperature (Hobden and Russell, 523 em7l establishes the optic phonon energy at 1964) the structure was divided into 3 sections. the center of the zone and is in good agreement with a. 782 - 718 cm 1: summation bands due to the value deduced from neutron scattering and processes involving creation of pairs of optic pho- infrared absorption. nons. The experimental arrangement has already been b. 603 - 454 cm-1: summation bands due to described in § 2. It is likely that with the high processes involving creation of pairs of optic and powered lasers at present being developed and in acoustic phonons. the ion lasers that the I c. 325 - 185 cmw : difference bands particular argon Raman g j. involving 624 creation of an optic phonon and destruction of an a coustic phonon. From this division the structure observed was interpreted in terms of density of states maxima near the points L and X in the Brillouin zone. The assignment was self consistent but not unique, and could be considerably improved by measurements at low temperatures where no difference bands exist. We have recently completed measurements of the Raman spectrum of GaP at 20 OK, our analysis is not yet complete, however several conclusions can be made. There has also been an improvement in the detail visible in the spectra, particularly in the contribution due to the optic branches. We hope that this will enable us to make a critical point, analysis of the optic phonon spectra of GaP : to do this a lattice dynamic calculation will be necessary ; when this is completed we hope to report FIG. 4. our analysis. At 20 OK the second order spectrum again divides into 3 groups (fig. 4) and as before the analysis is (Cochran et al., 1961) shows that the LO and TO based on this division. branches cross so that although LO > TO at the a. 800 - 670 cm-l : summation bands due to zone center, fi0 > LO at the zone edge. The LA creation of pairs of optic phonons. branch is well separated from the TA, it almost b. 613 - 293 cm.-1: summation bands due to reaches the optic branches at the edge of the zone. pairs of optic and acoustic phonons and pairs of It is expected that the dispersion curves for GaP acoustic phonons. will be similar in general to the Ga As curves but c. 289 - 150 crri 1: summation bands due to not in detail. pairs of acoustic phonons from the TA branch. The edge at 289 em-1 is probably 2 x the top The three groups are terminated at high energies of the fiA branch, the edge at 548 em-1 is due to by well defined edges at 289 cm-1, 548 cm-1 and the creation of pairs of optic and acoustic phonons, 804 cm-1. A calculation for gallium arsenide the acoustic phonons coming from the top of the

TABLE 2 625

TABLE 2 (continued)

fiA branch. The edge at 804 cm’1 fixes the top of the 2 phonon density of states. The top of the Discussion 2 phonon density of states may be due to 2 X LO (r) ; the gross structure at 786 cm-1 indi- M. TAYLOR. - What was the angle of incidence cates that the TO branch has moved up consi- of the exciting light in the silicon experiment ? derably when it reaches the edge of the zone an 8v,. RUSSELL. - About 450. in certain regions it may be greater than the value M. BiRMAN. -1) Have the relative intensities of LO at r. of t, vs. 2, - plionon Raman scattering processes The optic phonon energies at the center of the been determined for the three types of compounds : zone are established by the first order spectrum. covalent (e.g. C, Si), partially covalent (e.g. GaP, We find that ZnS, ... ), ionic (... ) ? If so, is there any evident trend ? If I that such absolute and = not, suggest LO (r) = 402 cm7l, TO (r) 366 cmm, relative intensity measurements would he most the relative intensities are LO = 3 X TO, the line valuable for the theory. widths are LO = 2 eni7l, fi0 = 4 cm-1. The LO 2) May I point out the necessity to measure expe- line is symmetrical about the centre ; but the TO rimentally the Raman scattering of circularly pola- line is asymetric on the low energy side. This rized light, i.e. circula.rly polarized incident light indicates that obscured on the low energy side of should be used, and the component of circularly the fi0 line there is a considerable two phonon den- polarized scattered light should be measured for sity of states leading to a reduction in the phonon either linear or circularly polarized incident light. lifetime. I shall point out in my paper on polarization (see The assignment in terms of density of states abstract this meeting), that it is essential to mea- maxima is given in Table 2. The assignments in sure the complete set of Stokes parameters of the terms of broad density of states maxima (Column 3) scattered light in order to completely characterize are reliable, but the detailed assignments in terms the scattering processes. of L,X and W much less certain, as they are based 3) In the same spirit as comment 2) above, I on detail in the curves for Ga As which suggest that the ruultipolarity (electric dipole, dispersion ’ may not apply to GaP. magnetic dipole, quadrupole, etc...) of the scattered radiation he measured to verify the assump- Acknowledgements. - We would like to thank tion always made (implicitly or explicitly) that the Dr. R. Loudon and Dr. F. A. Johnson for their Raman scattered radiation is electric dipole. Par- encouragement and advice throughout the work ticularly this may be of importance in deciding described here, Mr. G. Constant for his assis- about Ranian-lat.-tice scattering due to spin-flip tance during all the experiments, Mr. S. Brooks, type processes as I propose in my talk on theory Mrs. J. Dady and Mrs. W. A. Lavin for preparing of Raman effect (see abstract below). all the samples we have measured. 4) Now that scattering due to 3-phonon pro- 626 cesses can be observed (due to tho beautiful expe- tion. It is essentially a double (dispersion) spec- rimental work of Dr. Russell) may I point out trometer utilising gratings and photon counting that in my 1963 paper (Ph ys. Rev. (1963)) I sugges- techniques. Great care was taken to redace scat- ted a direct optical check of phonon assign- tering by mirror surfaces, etc... The purpose of ments in diamond/zincblende was possible since the double spectrometer is to further diminish the certain 3-phonon processes were allowed (IR or amount of scattered light reaching the detector. Raman) and that their observations would make Other features are increased signal to noise at low unambiguous assignments at X and L point,s. light levels, and linearity in the reading of inten- hl. GlORDlBfAI:NE. - How difficult is it to main- sities. The equipment has taken several years to tain the 20 OK temperature of GaP in the high flux design and build and we have high hopes that it at the focus of the laser beam ? A very small will meet expectations. 1 am sorry that Dr. Filler absorption constant would lead to a significant is not here to describe his equipment in person. temperature rise. However I am sure that we will be hearing about M. BURSTEIN. - I believe that it may be of his researches in the not too distant future. interest to mention the Raman spectrograph which M. MARADUDIN. - What is the possibility of Dr. Filler has designed and build at the University making an absolute measurement of the intensity of Pennsylvania and which will shortly be in opera- of the first order Raman spectrum of silicon ?

REFERENCES

BARKER (A. S.), Phys. Rev., 1964, 135, A 740-A 747. HOBDEN (M. V.) and RUSSELL (J. P.), Phys. Letters, 1964, COCHRAN (W.), FRAY (S. J.), JOHNSON (F. A.), QUARRING- 13, 39-41. TON (J. E.) and WILLIAMS (N.), J. App. Phys., Supp., RAMAN (Sir C. V.), Ind. J. Phys., 1928, 2, 387. 1960, 32, 2102. RUSSELL (J. P.) and LOUDON (R.), Proc. Phys. Soc., 1965, ELLIOTT (R. J.) and LOUDON (R.), Phys. Letters, 1963, 3, 85,1029-1033. 189-191. RUSSELL (J. P.), Proc. Phys. Soc., 1965, 85, 194-196. RUSSELL (J. P.), Appl. Phys. Letters, 1965, June 1.

fAPPENDIX ADDED IN PROOF

Recent work [1], has shown that the spectrum (3) Measurements at low temperatures (77 oK and believed to be the second order Raman spectrum of 20 OK) have shown remarkable increases in the inten- calcium fluoride is in fact a fluorescence due to Er3+. sity of the spectrum, these are incompatible with the In some recent work in Malvern we have satisfied expected temperature dependence for second order ourselves that the spectrum is not a second order Raman scattering. Raman spectrum. Although we have not identified (4) Prolonged exposures using the resonance mer- the Er3+ ions as the cause of the spectrum, our obser- cury line on the same crystal which we used have not vations are in general agreement with those of Gee, shown the spectrum [2]. In conclusion, there are two O’Shea and Cummins ; the observations which we features of the spectrum which remain puzzling ; have made are listed below : (1) As Gee, O"Shea and Cummins point out Car2 : (1) Although the spectrum has been observed in a Er3+ is not known to absorb at 6328 A. number of different crystals we have now obtained (2) The extent of the spectrum is in reasonable some crystals which do not show it. agreement with that which would be expected for the (2) Prolonged exposures (up to 64 hours) on those second order Raman spectrum from the known dis- crystals which show the spectrum have failed to show persion curves, and of course the second order Raman anything other than the first order line on the anti- spectrum of calcium fluoride has still to be determined. stokes side.

REFERENCES [1] GEE (A. R.), O’SHEA (D. C.) and CUMMINS (H. Z.), (in Press). [2] KRISHNAN (R. S.), 1965, Private Communication.