- Email: [email protected]
Surface structure analysis using low energy scattered and recoiling ions
Nuclear Instruments and Methods North-Holland. Amsterdam
SURFACE STRUCTURE RECOILING IONS D.J. O’CONNOR,
in Physics
ANALYSIS
Research
B13 (1986)
USING
R.J. MACDONALD,
235-239
235
LOW ENERGY
W. ECKSTEIN”
and
SCATTERED
AND
P.R. HIGGINBOTTOM
Physics Department, University of Newcastle, NSW 2308 Australia “Max Planck Institut fiir Plasmaphysik, Garching, Munich, West Germany
There have been many studies of surface structure and adsorption using low energy ion scattering (LEIS). A well quoted example is the study of oxygen adsorption on the Ni(ll0) surface, but despite the numerous studies there is not yet agreement on the adsorption site. It is evident that LEIS alone cannot yield enough information to differentiate between some of the proposed models and LEED does not provide the necessary complementary information required to distinguish between similar models. In a new study additional information has been gathered by combining the results of LEIS with low energy hydrogen scattering and low energy negative ion recoil spectroscopy. The hydrogen scattering experiments yield similar information to its medium energy and high energy (RBS) analogues by the processes of channeling and blocking to locate the adsorption site relative to the underlying atoms. The low energy negative ion recoil process is extremely sensitive to the electronegative atoms and is thus a perfectly tailored tool for oxygen studies. The angular distributions of the negative ions yield valuable information concerning its initial physical adsorption site. These three aspects of low energy ion bombardment have been applied to the Ni(llO)-0 surface and the agreement with various proposed surface structures is discussed.
1. Introduction Low energy ion scattering using inert gas and alkali ions is widely used in studies of the surface atomic layer. The extreme surface sensitivity of this technique ensures that it yields both compositional and structural information on clean and adsorbate covered surfaces. A number of investigations have been carried out on initial adsorbate site location using inert gas ion LEIS [l-4]. The conclusions usually invoke blocking or shadowing of substrate atoms to explain changes in reflected ion yield though attention is rarely paid to possible problems arising from variations in the charge exchange process [4]. It is possible to substantiate the structures determined by LEIS by using slight variations of the technique which yield substantially different information. These variations involve the use of H’ in the same channelling and blocking geometries as those developed for medium and high energy ion scattering, and the use of the highly sensitive negatively charged recoils for adsorbates which have a strong chance of being found in the negative charged state (e.g. oxygen). The use of channelling and blocking is well documented [5-71 for ion energies greater than 50 keV and there have been some reported uses at low energies [S-lo], however it has not been regularly used at low energies for surface analysis. At low energies (~10 keV) the size of the shadow cone for H’ projectiles is much larger than the thermal vibrations of the atoms in the atomic strings thus simplifying the inter0168-583X/86/$03.50 0 Elsevier Science Publishers (North-Holland Physics Publishing Division)
B.V
pretation of measurements. Only the first atom of each string will effectively contribute to the “surface peak” of a channelled distribution. Thus the surface peak of the hydrogen channelling spectrum at low energy is a true surface peak. Weighed against this advantage must be the realisation that charge exchange processes play an important role in determining the scattered ion yield at low energies and thus must also be considered when discussing changes in ion yield. In general the detection of recoil ions is difficult in the positive spectrum as they may correspond to energies at which there is a strong scattered ion signal, or they may lie on the tail of a scattered ion signal. These problems can be avoided by monitoring the negatively charged particles. The advantages associated with the use of negatively charged recoils are twofold: (a) Being negatively charged they do not need to be extracted from an intense background of positively charged scattered ions if inert gas primary projectiles are used as the probability of creating negatively charged inert gas ions is exceedingly small. of ejected adsorbate (b) The angular distribution atoms carries with it direct information about the site from which ejection occurred. This is a much more direct measurement of the adsorption site than that obtained by inference from changes in scattered ion yield. Both H‘ scattering and negatively charged recoils have been applied to a study of oxygen on Ni (110) to gauge the sensitivity to coverage and site location. VI. LOW
ENERGY
SURFACE
INTERACTIONS
236
D. J. O’Connor
el al. I Surface structure analysis
2. Experimental
The experimental system incorporates a 5 keV 3M ion gun, a specimen manipulator with two rotational and three translational degrees of freedom and an electrostatic energy analyser with two rotational degrees of freedom. The vacuum chamber is pumped by a combination of turbomolecular and titanium sublimation pumps which achieve a base pressure of 5 x lo-” mbar. In the experiments involving hydrogen scattering 2 keV Hl were used. The target was a 99.999% pure Ni single crystal aligned to within 0.5” of a (110) surface. The target was mechanically polished and final cleaning was achieved by sputter bombardment with 2 keV Ne’ at 700 K. The samples revealed no evidence of contamination to the limits of measurement using LEIS (0.01 of a monolayer) though the use of negatively charged ions is so much more sensitive than observing positively charged scattered ions that it was found that regardless of how apparently clean a surface is when monitored with conventional LEIS, there is always a significant signal of negatively charged recoil ions. This attests to the extreme sensitivity of the negative ions to low levels of adsorbates. Rough estimates suggest the negative ion yield will allow monitoring of surface oxygen to about 10m5 monlayers or less. The adsorption procedure followed was aimed at reproducing conditions reported in similar LEIS and LEED studies of oxygen on Ni(ll0) to yield a (2 X 1) LEED pattern. After target cleaning and annealing of damage the sample was exposed to 0.4 Langmuirs (L) of oxygen at a sample temperature of 450 K. The measurements were performed with a beam dose which was experimentally found to produce minimal changes to the surface structure by sputtering or surface damage. After each measurement the target was cleaned with sputter bombardment at 700 K.
3. Results and discussion The surface structure and alignment can be verified using the channelling and blocking features of the H’ scattering, as shown in fig. la. This illustrates scattering from a clean surface. In this figure the energy spectra (over the range 0.8 < EIE,,
channelling and blocking associated with the more closely packed directions. The adsoprtion of 0.4 L of oxygen results in a marked decrease in the height of the surface peak when the ion beam is directed along the (010) surface direction and the detector is aligned with the (100) direction. The change in the form of the channelling structure with the adsorption of oxygen is shown in fig. lb. It could equally well be argued from this information that the oxygen sat in the short bridge site shielding the surface atomic layer or that it sat in the long bridge site shielding the second atomic layer sites. To resolve this ambiguity with hydrogen ion scattering alone it is necessary to identify site specific alignment directions. An energy spectrum of the negatively charged particles emitted from the surface (fig. 2) comprises a relatively narrow low energy peak which results from a single binary event creating a direct recoil particle. There is often observed one or two higher energy peaks in the negative particle energy spectrum which have been identified as originating from deflected recoil processes. The term ‘deflected recoil’ is collectively given to the processes which involve scattering of an atom after the initial recoil event, and to the recoil event where the projectile has previously been scattered by another atom. There is no general rule to distinguish these two processes, and they must therefore be differentiated by reference to computer simulations. An effective demonstration of the usefulness of recoil ions is revealed in fig. 3. The recoiling O- is shown for a recoil angle of 60” while the angle of incidence is increased from 0” (measured to the surface) up to 60” when a 2 keV Ne’ projectile is used. The magnitude of the direct recoil signal exhibits marked differences between the case when the experiment is performed along the (001) surface chain and the (110) surface chain. These differences can be used to distinguish between different adsorption sites. It is possible to call on previous extensive studies of this surface or the currently accepted reconstruction model to eliminate some the suggested adsorption sites. However it is instructive to apply LENRS to this surface without assuming any a priori knowledge of the surface structure to illustrate the potential application of this technique to as yet unstudied surfaces. The results in fig. 3 preclude the possibility that the oxygen is sitting in a centrefold site. If this were so the recoil signals would be comparable along the two surface alignments. The short bridge site (between atoms in the (001) surface direction) is also ruled out as that would yield a higher signal along the (001) surface direction. The only oxygen position consistent with these results is a long bridge site (between atoms in the (001) surface direction). Under the experimental conditions used in this study the projectile is prevented from striking the
D..i.
10.
15.
20.
2s.
O’Connor
et al. i Surface
30.
3s.
sfmdure
PO,
analysis
bS.
237
SO.
ANGLE OF INCIDENCE
355.
300.
250.
ZOO. D ;: ;:
ISO.
150.
SO.
10.
15.
262.
2s. 30. ANGLE OF INCIDENCE
35.
ID.
15.
SO.
Fig. 1. Hydrogen scattering spectra from clean (a) and oxygen adsorbed (b) Ni( 110) surface when the ion beam is aligned with the (010) direction and the detector is aligned with the (l(K)) direction. Each energy spectrum is taken over the energy range (0.8< E/E, < I.O>.
oxygen along the (001) direction because it lies in the shadow cones formed by the neighbouring Ni atoms. Using the Molikre function with an appropriate screening length correction to predict the shadow cone dimensions it is possible to determine that the oxygen must be more than 0.42 A above the Ni surface in order that a comparable yield to the ( 110) direction is seen along the (001) direction under these experimental conditions. The results in fig. 3 indicate that the oxygen is
closer than 0.42 A to the surface. In a more extensive study of recoil ions [ll] it has been shown that the height of the oxygen above the surface is 0.2 2 0.1 A. This height results in a NCCl bond length of 1.77 J% which is in close agreement with the value of 1.75~% predicted by Mitchell 1121. There have been recent reports af similar findings for oxygen on Ni determined by other ion scattering techniques [13, 141. While there exists a wealth of structural information VI. LOW ENERGY SURFACE INTERACTIONS
D. _I. O’Connor
238
et al.
I Surface
structure
analysis
250.
1
.30
.40
.so
-60
-70
-60
-90
4
E / EO
Fig. 2. A negative ion recoil spectrum showing the direct recoil component is 2 keV Ne’ and the recoil gngle is 60”. -
(R) and two deflected
recoil peaks (D). The incident
ion
in the angular distribution of the directly recoiling atoms there is a greater potential for the information to be extracted from the deflected recoils. As these events involve both a recoiling event and a scattering event they may be used to determine the relative positions of the Ni and 0 atoms. This aspect requires more attention and comparison to computer simulation.
4. Conclusions The use of hydrogen scattering and negatively charged recoil ions greatly extends the capabilities of low energy sattering techniques. In particular the negatively charged recoils is extremely sensitive to electronegative elements. These methods have been successfully applied to a study of 0 on Ni( 110) and located the adsorbate in the long bridge site just above the surface. Additional work is in progress to assess the suitability of the deflected recoils to more detailed structural information.
References 0
10
20 ANGLE
30
40
50
60
OF INCIDENCE
Fig. 3. The direct O- recoil yield as a function of the angle of incidence along the (100) and (110) surface low index directions for a fixed recoil angle of 60”. This yield dependence identifies the oxygen position as a long bridge site. The possible adsorption sites are shown in the insert. The surface normal projection identifies the Ni atoms by closed circles and the open circles labelled L, S and C refer to the long bridge, the short bridge and the centrefold sites respectively.
Surf. Sci. 68 (1977) 96. [II W. Heiland and E. Taglauer, [21 L.K. Verheij, J.A. Van Den Berg and D.G. Armour, Surf. Sci. 84 (1979) 408. [31 J.A. Van Den Berg, L.K. Verheij and D.G. Armour, Surf. Sci. 91 (1980) 218. [41 W. Englert, E. Taglauer and W. Heiland, Surf. Sci. 117 (1982) 124. W. Soska, F.W. Saris, H.H. Kersten [51 W.C. Turkenburg, and B.C. Colenbrander, Nucl. Instr. and Meth. 132 (1976) 587 and refs. contained therein.
D.J.
O’Connor
et al. I Surface structure analysis
[6] R.G. Smeenk, R.M. Tromp. J.F. Van der Veen and F.W. Saris, Surf. Sci. 95 (1980) 156. [7] L.C. Feldman, Nucl. Instr. and Meth. 191 (1981) 211. [8] H.H.W. Feijen, L.K. Verhey and A.L. Boers, Phys. Lett. 45A (1973) 31. [9] H.H.W. Feijen, E.P.Th.M. Suurmeijer and A.L. Boers, Atomic Collisions in Solids, vol 2 (1975) p. 573.
[lo] [11] [12] [13] [14]
239
H.H.W. Feijen, Thesis, Groningen (1975). D.J. O’Connor, submitted to Surf. Sci. K.A.R. Mitchell, Surf. Sci. 92 (1980) 79. M. Schuster and C. Varelas, Surf. Sci. 134 (1983) 195. H. Niehus and G. Comsa, Surf. Sci. 151 (1985) 1171.
VI. LOW
ENERGY
SURFACE
INTERACTIONS
SURFACE STRUCTURE RECOILING IONS D.J. O’CONNOR,
in Physics
ANALYSIS
Research
B13 (1986)
USING
R.J. MACDONALD,
235-239
235
LOW ENERGY
W. ECKSTEIN”
and
SCATTERED
AND
P.R. HIGGINBOTTOM
Physics Department, University of Newcastle, NSW 2308 Australia “Max Planck Institut fiir Plasmaphysik, Garching, Munich, West Germany
There have been many studies of surface structure and adsorption using low energy ion scattering (LEIS). A well quoted example is the study of oxygen adsorption on the Ni(ll0) surface, but despite the numerous studies there is not yet agreement on the adsorption site. It is evident that LEIS alone cannot yield enough information to differentiate between some of the proposed models and LEED does not provide the necessary complementary information required to distinguish between similar models. In a new study additional information has been gathered by combining the results of LEIS with low energy hydrogen scattering and low energy negative ion recoil spectroscopy. The hydrogen scattering experiments yield similar information to its medium energy and high energy (RBS) analogues by the processes of channeling and blocking to locate the adsorption site relative to the underlying atoms. The low energy negative ion recoil process is extremely sensitive to the electronegative atoms and is thus a perfectly tailored tool for oxygen studies. The angular distributions of the negative ions yield valuable information concerning its initial physical adsorption site. These three aspects of low energy ion bombardment have been applied to the Ni(llO)-0 surface and the agreement with various proposed surface structures is discussed.
1. Introduction Low energy ion scattering using inert gas and alkali ions is widely used in studies of the surface atomic layer. The extreme surface sensitivity of this technique ensures that it yields both compositional and structural information on clean and adsorbate covered surfaces. A number of investigations have been carried out on initial adsorbate site location using inert gas ion LEIS [l-4]. The conclusions usually invoke blocking or shadowing of substrate atoms to explain changes in reflected ion yield though attention is rarely paid to possible problems arising from variations in the charge exchange process [4]. It is possible to substantiate the structures determined by LEIS by using slight variations of the technique which yield substantially different information. These variations involve the use of H’ in the same channelling and blocking geometries as those developed for medium and high energy ion scattering, and the use of the highly sensitive negatively charged recoils for adsorbates which have a strong chance of being found in the negative charged state (e.g. oxygen). The use of channelling and blocking is well documented [5-71 for ion energies greater than 50 keV and there have been some reported uses at low energies [S-lo], however it has not been regularly used at low energies for surface analysis. At low energies (~10 keV) the size of the shadow cone for H’ projectiles is much larger than the thermal vibrations of the atoms in the atomic strings thus simplifying the inter0168-583X/86/$03.50 0 Elsevier Science Publishers (North-Holland Physics Publishing Division)
B.V
pretation of measurements. Only the first atom of each string will effectively contribute to the “surface peak” of a channelled distribution. Thus the surface peak of the hydrogen channelling spectrum at low energy is a true surface peak. Weighed against this advantage must be the realisation that charge exchange processes play an important role in determining the scattered ion yield at low energies and thus must also be considered when discussing changes in ion yield. In general the detection of recoil ions is difficult in the positive spectrum as they may correspond to energies at which there is a strong scattered ion signal, or they may lie on the tail of a scattered ion signal. These problems can be avoided by monitoring the negatively charged particles. The advantages associated with the use of negatively charged recoils are twofold: (a) Being negatively charged they do not need to be extracted from an intense background of positively charged scattered ions if inert gas primary projectiles are used as the probability of creating negatively charged inert gas ions is exceedingly small. of ejected adsorbate (b) The angular distribution atoms carries with it direct information about the site from which ejection occurred. This is a much more direct measurement of the adsorption site than that obtained by inference from changes in scattered ion yield. Both H‘ scattering and negatively charged recoils have been applied to a study of oxygen on Ni (110) to gauge the sensitivity to coverage and site location. VI. LOW
ENERGY
SURFACE
INTERACTIONS
236
D. J. O’Connor
el al. I Surface structure analysis
2. Experimental
The experimental system incorporates a 5 keV 3M ion gun, a specimen manipulator with two rotational and three translational degrees of freedom and an electrostatic energy analyser with two rotational degrees of freedom. The vacuum chamber is pumped by a combination of turbomolecular and titanium sublimation pumps which achieve a base pressure of 5 x lo-” mbar. In the experiments involving hydrogen scattering 2 keV Hl were used. The target was a 99.999% pure Ni single crystal aligned to within 0.5” of a (110) surface. The target was mechanically polished and final cleaning was achieved by sputter bombardment with 2 keV Ne’ at 700 K. The samples revealed no evidence of contamination to the limits of measurement using LEIS (0.01 of a monolayer) though the use of negatively charged ions is so much more sensitive than observing positively charged scattered ions that it was found that regardless of how apparently clean a surface is when monitored with conventional LEIS, there is always a significant signal of negatively charged recoil ions. This attests to the extreme sensitivity of the negative ions to low levels of adsorbates. Rough estimates suggest the negative ion yield will allow monitoring of surface oxygen to about 10m5 monlayers or less. The adsorption procedure followed was aimed at reproducing conditions reported in similar LEIS and LEED studies of oxygen on Ni(ll0) to yield a (2 X 1) LEED pattern. After target cleaning and annealing of damage the sample was exposed to 0.4 Langmuirs (L) of oxygen at a sample temperature of 450 K. The measurements were performed with a beam dose which was experimentally found to produce minimal changes to the surface structure by sputtering or surface damage. After each measurement the target was cleaned with sputter bombardment at 700 K.
3. Results and discussion The surface structure and alignment can be verified using the channelling and blocking features of the H’ scattering, as shown in fig. la. This illustrates scattering from a clean surface. In this figure the energy spectra (over the range 0.8 < EIE,,
channelling and blocking associated with the more closely packed directions. The adsoprtion of 0.4 L of oxygen results in a marked decrease in the height of the surface peak when the ion beam is directed along the (010) surface direction and the detector is aligned with the (100) direction. The change in the form of the channelling structure with the adsorption of oxygen is shown in fig. lb. It could equally well be argued from this information that the oxygen sat in the short bridge site shielding the surface atomic layer or that it sat in the long bridge site shielding the second atomic layer sites. To resolve this ambiguity with hydrogen ion scattering alone it is necessary to identify site specific alignment directions. An energy spectrum of the negatively charged particles emitted from the surface (fig. 2) comprises a relatively narrow low energy peak which results from a single binary event creating a direct recoil particle. There is often observed one or two higher energy peaks in the negative particle energy spectrum which have been identified as originating from deflected recoil processes. The term ‘deflected recoil’ is collectively given to the processes which involve scattering of an atom after the initial recoil event, and to the recoil event where the projectile has previously been scattered by another atom. There is no general rule to distinguish these two processes, and they must therefore be differentiated by reference to computer simulations. An effective demonstration of the usefulness of recoil ions is revealed in fig. 3. The recoiling O- is shown for a recoil angle of 60” while the angle of incidence is increased from 0” (measured to the surface) up to 60” when a 2 keV Ne’ projectile is used. The magnitude of the direct recoil signal exhibits marked differences between the case when the experiment is performed along the (001) surface chain and the (110) surface chain. These differences can be used to distinguish between different adsorption sites. It is possible to call on previous extensive studies of this surface or the currently accepted reconstruction model to eliminate some the suggested adsorption sites. However it is instructive to apply LENRS to this surface without assuming any a priori knowledge of the surface structure to illustrate the potential application of this technique to as yet unstudied surfaces. The results in fig. 3 preclude the possibility that the oxygen is sitting in a centrefold site. If this were so the recoil signals would be comparable along the two surface alignments. The short bridge site (between atoms in the (001) surface direction) is also ruled out as that would yield a higher signal along the (001) surface direction. The only oxygen position consistent with these results is a long bridge site (between atoms in the (001) surface direction). Under the experimental conditions used in this study the projectile is prevented from striking the
D..i.
10.
15.
20.
2s.
O’Connor
et al. i Surface
30.
3s.
sfmdure
PO,
analysis
bS.
237
SO.
ANGLE OF INCIDENCE
355.
300.
250.
ZOO. D ;: ;:
ISO.
150.
SO.
10.
15.
262.
2s. 30. ANGLE OF INCIDENCE
35.
ID.
15.
SO.
Fig. 1. Hydrogen scattering spectra from clean (a) and oxygen adsorbed (b) Ni( 110) surface when the ion beam is aligned with the (010) direction and the detector is aligned with the (l(K)) direction. Each energy spectrum is taken over the energy range (0.8< E/E, < I.O>.
oxygen along the (001) direction because it lies in the shadow cones formed by the neighbouring Ni atoms. Using the Molikre function with an appropriate screening length correction to predict the shadow cone dimensions it is possible to determine that the oxygen must be more than 0.42 A above the Ni surface in order that a comparable yield to the ( 110) direction is seen along the (001) direction under these experimental conditions. The results in fig. 3 indicate that the oxygen is
closer than 0.42 A to the surface. In a more extensive study of recoil ions [ll] it has been shown that the height of the oxygen above the surface is 0.2 2 0.1 A. This height results in a NCCl bond length of 1.77 J% which is in close agreement with the value of 1.75~% predicted by Mitchell 1121. There have been recent reports af similar findings for oxygen on Ni determined by other ion scattering techniques [13, 141. While there exists a wealth of structural information VI. LOW ENERGY SURFACE INTERACTIONS
D. _I. O’Connor
238
et al.
I Surface
structure
analysis
250.
1
.30
.40
.so
-60
-70
-60
-90
4
E / EO
Fig. 2. A negative ion recoil spectrum showing the direct recoil component is 2 keV Ne’ and the recoil gngle is 60”. -
(R) and two deflected
recoil peaks (D). The incident
ion
in the angular distribution of the directly recoiling atoms there is a greater potential for the information to be extracted from the deflected recoils. As these events involve both a recoiling event and a scattering event they may be used to determine the relative positions of the Ni and 0 atoms. This aspect requires more attention and comparison to computer simulation.
4. Conclusions The use of hydrogen scattering and negatively charged recoil ions greatly extends the capabilities of low energy sattering techniques. In particular the negatively charged recoils is extremely sensitive to electronegative elements. These methods have been successfully applied to a study of 0 on Ni( 110) and located the adsorbate in the long bridge site just above the surface. Additional work is in progress to assess the suitability of the deflected recoils to more detailed structural information.
References 0
10
20 ANGLE
30
40
50
60
OF INCIDENCE
Fig. 3. The direct O- recoil yield as a function of the angle of incidence along the (100) and (110) surface low index directions for a fixed recoil angle of 60”. This yield dependence identifies the oxygen position as a long bridge site. The possible adsorption sites are shown in the insert. The surface normal projection identifies the Ni atoms by closed circles and the open circles labelled L, S and C refer to the long bridge, the short bridge and the centrefold sites respectively.
Surf. Sci. 68 (1977) 96. [II W. Heiland and E. Taglauer, [21 L.K. Verheij, J.A. Van Den Berg and D.G. Armour, Surf. Sci. 84 (1979) 408. [31 J.A. Van Den Berg, L.K. Verheij and D.G. Armour, Surf. Sci. 91 (1980) 218. [41 W. Englert, E. Taglauer and W. Heiland, Surf. Sci. 117 (1982) 124. W. Soska, F.W. Saris, H.H. Kersten [51 W.C. Turkenburg, and B.C. Colenbrander, Nucl. Instr. and Meth. 132 (1976) 587 and refs. contained therein.
D.J.
O’Connor
et al. I Surface structure analysis
[6] R.G. Smeenk, R.M. Tromp. J.F. Van der Veen and F.W. Saris, Surf. Sci. 95 (1980) 156. [7] L.C. Feldman, Nucl. Instr. and Meth. 191 (1981) 211. [8] H.H.W. Feijen, L.K. Verhey and A.L. Boers, Phys. Lett. 45A (1973) 31. [9] H.H.W. Feijen, E.P.Th.M. Suurmeijer and A.L. Boers, Atomic Collisions in Solids, vol 2 (1975) p. 573.
[lo] [11] [12] [13] [14]
239
H.H.W. Feijen, Thesis, Groningen (1975). D.J. O’Connor, submitted to Surf. Sci. K.A.R. Mitchell, Surf. Sci. 92 (1980) 79. M. Schuster and C. Varelas, Surf. Sci. 134 (1983) 195. H. Niehus and G. Comsa, Surf. Sci. 151 (1985) 1171.
VI. LOW
ENERGY
SURFACE
INTERACTIONS