Interface evolution of Co/Ti multilayers with ultra-short period P. Sarkar , A. Biswas , S. Ghosh , S. Rai , M.H. Modi , D. Bhattacharyya PII: DOI: Reference:
S0040-6090(19)30715-1 https://doi.org/10.1016/j.tsf.2019.137688 TSF 137688
To appear in:
Thin Solid Films
Received date: Revised date: Accepted date:
18 April 2019 4 November 2019 4 November 2019
Please cite this article as: P. Sarkar , A. Biswas , S. Ghosh , S. Rai , M.H. Modi , D. Bhattacharyya , Interface evolution of Co/Ti multilayers with ultra-short period, Thin Solid Films (2019), doi: https://doi.org/10.1016/j.tsf.2019.137688
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier B.V.
Co/Ti multilayers with 18Å bi-layer thickness deposited by magnetron sputtering.
Multilayers with 10,20,40,60 and 80 bi-layers have been grown.
Samples characterized by specular and diffused hard X-ray reflectivity measurements.
Interface diffusion of Co into Ti layers significant in these multilayers.
Interface roughness does not increase with number of bi-layers.
Interface evolution of Co/Ti multilayers with ultra-short period P.Sarkar1,2, A. Biswas1, S. Ghosh3, S. Rai4, M.H. Modi4 and D. Bhattacharyya1,* 1
Atomic & Molecular Physics Division, Bhabha Atomic Research Centre, Mumbai-400 085, India Homi Bhabha National Institute, Mumbai-400 094, India 3 Material Processing Division, Bhabha Atomic Research Centre, Mumbai-400 085, India 4 Synchrotron Utilisation Section, Raja Ramnna Centre for Advanced Technology, Indore- 752013, India 2
Abstract Behaviour of interface properties and propagation of interfacial atomic diffusion have been investigated systematically in Co/Ti multilayers with ultra-short (~18Å) bi-layer thickness. Co/Ti multilayers with 10,20,40,60 and 80 bi-layers have been grown using an indigenously built inline dc magnetron sputtering system having application as reflecting optics in the water window soft X-ray regime of 23–44 Å. Specular and diffused Grazing Incidence X-ray reflectivity measurements with hard X-rays have been used to evaluate detailed interfacial structure of the multilayers. It is observed that interfacial atomic diffusion of Co into Ti layers is significant in such ultra-short period multilayer structures and it increases as more number of layers are added in the multilayer stack. As a result of Co diffusion into Ti the physical density of Ti layer increases and also effective thickness of Co layer decreases. Since the thickness of the Ti layers are in sub-nm range, it is found that this diffusion is not limited only to the interface region of Co-on-Ti, rather it spreads throughout the bulk of Ti layer. Finally, soft X-ray reflectivity measurement on the multilayers with synchrotron radiation at 28 Å wavelength shows that the peak reflectivity is limited by the significant diffusion at the Co/Ti interfaces.
Key words: Magnetron Sputtering, Water Window, Grazing Incidence X-ray Reflectivity *corresponding author (email: [email protected]
1. Introduction Advancements in the fields of instrumentations in extreme ultra-violet and soft X-ray based lithography and microscopy , hard X-ray telescopes  and gamma-ray astronomy  have scaled down the characteristic length-scale of X-ray optics to a few nanometers to subnanometer range. Soft X-ray microscopy with Schwarzschild type objective  in the waterwindow region (23–44 Å) is one of these important fields, where X-ray multilayer mirrors play major role. Because of very short wavelength application, each layer thickness of such multilayers should be in the sub-nm range for near normal incidence. The properties of such Xray multilayer structures are strongly influenced by the layer thickness, total number of layers, thickness ratio of two components, the interfacial interaction and microstructure. Among these, interfacial atomic diffusion and interface roughness play most crucial role in controlling the physical properties of the structures. Even the best-compatible material combinations (according to the optical, chemical, and physical parameters) exhibit poor reflectivity in reality. This may be because of accumulated roughness propagation  and/or intermixing at interface  resulting in the lack of formation of flat and abrupt interfaces . Therefore, future advancement in such nano scaled multilayer based system requires detailed scientific understanding of the interface properties. In soft X-ray water-window regime, transition metals such as Sc, Ti, V, and Ni can be used as spacer material for developing near normal incidence mirrors by utilizing anomalies in refractive indices at their respective absorption edges. Several reports are available in literature on different material combinations such as Cr/Sc , TiO2/Al2O3 , V/Sc , Ni/V , Ni/Ti , Fe/V  etc. for application as focusing optics in water-window regime range. However, there is not many reports on the combination of Co/Ti although this is one of the potentially good
candidates for water window soft X-ray regime applications above Ti absorption edge (~27.4 Å) according to theoretical reflectivity calculations . We have earlier reported  the results of Co/Ti multilayers having relatively higher (~42 Å) bilayer thickness to explore the feasibility of this multilayer system, as reflecting optics in soft X-ray water window region. The above study was motivated from our experience in producing very high quality Co/Ti multilayers for neutron supermirror  applications. In the above work , we have observed that on contrary to what had been observed in case of Co/Ti and many other multilayer systems having relatively higher bi-layer thickness, Co/Ti multilayers do not really follow the “restart of growth model” in this moderate bi-layer thickness regime (~42 Å). We have also demonstrated that in this case, the interfaces are dominated by interface roughness, and the cumulative interface roughness propagation occurs throughout the multilayer which limits the highest reflectivity obtained by the multilayer. In the above work, the Co/Ti multilayer produced showed significant reflectivity in the water-window region, but at a low grazing angle of incidence of 20 due to relatively higher bi-layer thickness used. There is another report on Co/Ti multilayer system in water-window regime by Zhu et al. where they have reported that soft X-ray reflectivity of a Co/Ti multilayer mirror can be considerably improved by the introduction of nitrogen gas in an optimised percentage during film growth . The above study is also on multilayers with relatively larger bi-layer thicknesses and lower number of bilayers (Nmax=20). However, since for application in water window microscope these multilayers are supposed to work at higher grazing angle of incidence (ideally at near normal incidence) in the present report we have prepared Co/Ti multilayers of much lower bi-layer thickness of ~18Å
(with nominally 9Å of Co and 9Å of Ti layers) to obtain reflectivity peak at a grazing angle of incidence of 40 or above. For such multilayers having individual layer thickness in the range of below 1 nm, where the Volmer-Weber growth mode dominates and most of the materials are near percolation transition regime, the interaction and diffusion processes at the interfaces are expected to differ significantly as compared to multilayers having higher thickness. In this communication, we have focused mainly on the interface roughness as well as on atomic diffusion evolution in Co/Ti multilayers in ultra-short thickness regime, with each layer thickness in sub-nm range, quantitatively and qualitatively investigated by using specular and diffused Grazing Incidence Xray reflectivity (GIXR) measurements.
2. Experimental details 2.1. Preparation of samples: The single layer Co and Ti films as well as Co/Ti multilayers have been deposited using the 9 m long indigenously developed in-line magnetron sputtering system described elsewhere . This system is equipped with three magnetrons on which three sputtering targets, each having dimensions of 125 mm × 200 mm are fixed. During the deposition process the substrate trolley is scanned below the targets with a stainless steel (SS) pulley rope mechanism, using a motor kept outside the vacuum system and the deposited film thickness is scaled with the inverse of the trolley velocity. During deposition, the online tracking of the exact position of the substrate trolley has been achieved by a laser beam reflection from the back of the trolley tray, the detail description of which has been given in our previous communication . Experimental realization of deposition of a layer having thickness as low as 10 Å, using
such in-line sputtering system, is itself a very challenging job. The minimum thickness value that can be achieved is limited mainly by two factors, viz., (i) the minimum power given to the magnetron at which plasma can be sustained without fluctuation and (ii) the maximum velocity of substrate trolley, without any jerking, which decides the minimum rate of deposition. With the minimum dc power level of 200 W (Co/ Ti) for stable and uninterrupted plasma and a maximum substrate trolley speed of 68 mm sec−1 for smooth movement of the trolley, the deposited layer is found to be ~20 Å, by a single pass of the substrate trolley. This is however, far above than the required thickness value. For further reduction of the rate of deposition, a specially designed mask, made of SS is used below the target at a distance of 5 cm. This mask reduces the exposed length of the target along the direction of the trolley movement and thus results in a drastic reduction of the rate of deposition, without interrupting the plasma. The mask is schematically shown in fig.1. As the mask covers the electron trap path onto magnetron, the rate of deposition gets reduced almost by a factor of 2. Prior to deposition, the system has been evacuated to a base pressure of
1×10−4 Pa with the help of two turbo-molecular pumps. High-quality Ti (99.95%)
and Co (99.95%) targets of specified dimensions are used for all of the depositions. All of the films have been deposited on 30 mm × 20 mm crystalline silicon (111) wafers.
2.2. Characterization of samples: Initially an optimization of the Ti and Co layers for ultrathin thickness is performed using the mask from single layer GIXR data analysis. Subsequently, a set of multilayer samples are prepared using DC power of 400 and 300 W for Ti and Co layers respectively with 10, 20, 40, 60 and 80 numbers of bi-layers at a nominal bi-layer thickness of ~18 Å (with nominal values of 9 Å of Co and 9 Å of Ti layer thickness). Thickness, density, and top surface roughness, as well as
interface width of the films and the multilayers, have been determined by fitting their GIXR data recorded in both specular and non-specular (diffused) modes. Under the Parrat formalism , the specular reflectivity of X-rays from a multilayer thin film with ideal sharp and plane interfaces, is obtained by applying Fresnels‟ equation at each interface. Analysis of this specular X-ray reflectivity spectrum can yield information regarding density and thickness of a thin film/multilayer accurately. However, in the case of non-abrupt interfaces, the reflectivity is modified as,
R R0 e q 2
is the momentum transfer factor, R is the reflectivity of the rough surface
and R0 is the reflectivity of an otherwise identical smooth surface, is the wavelength of X-ray, is the rms roughness of the surface. In case of an interface, the parameter is generally referred to as “interface width” and is a combined measure of interface roughness ( r ) and interface diffusion ( d ). These two contributions cannot be decoupled by measuring only specular reflectivity measurement and measurement of non-specular (diffuse) component of reflectivity is necessary in addition. Moreover, specular reflectivity represents the coherent part of entire reflected wave, which only includes the point properties of the rough surface i.e., its mean-square roughness. In order to investigate the in-plane correlation in a rough surface, measurement of incoherent component of reflectivity is necessary, which is represented by nonspecular reflectivity. GIXR spectra of all the samples discussed in this paper have been recorded both in specular and non-specular mode. The configuration of the above two scans in the reciprocal
space is shown in fig. 2, where z is the direction perpendicular to the sample surface. The two
components of the momentum transfer vector q ki k f are given by: q x k i (cos f cos i )
ki ( f i ) 2
q z k i (sin f sin i ) k i
is the scattering angle, i and f being the angles of incidence and reflections respectively measured from the sample surface. For specular reflectivity measurement, a standard ( 2 ) scan has been carried out at low grazing angle of incidence while diffused reflectivity has been measured in the “Rocking Scan” geometry. In this mode both the angle of incidence i and the reflection angle f are varied around peak position of the specular reflectivity spectrum in such a fashion that scattering angle i f remains same throughout whole scan. It can be seen from above that a specular scan ( i f ) means a q z scan with q x 0 while for rocking scan i and f are varied and is fixed, thus q z remains constant and q x component of the momentum transfer vector is varied. The measured diffused reflectivity is presented in terms of the quantity
1 dI ( ) where, I 0 d 0
dI is the intensity of X-rays scattered from the sample within a solid angle d 0 made by the
illuminated portion of the detector at the point of incidence on the sample surface (3*10 -5 Sr, in the present case, as defined by the slit width) and I 0 is the incident intensity.
The diffused reflectivity data have subsequently been analyzed by the formalism laid out by Windt et al. [17,18], based on the first order Distorted Wave Born Approximation (DWBA) where, each interface of a multilayer is described by a Power Spectral density (PSD) function. PSD function in turn depends on interface roughness ( r ), the in-plane or parallel correlation length ( || ) and the jaggedness factor ( h ). The measured diffused reflectivity data have been fitted with theoretical diffused reflectivity plots using the two fitting parameters for each interface r and || . Jaggedness parameter value has been kept fixed at h =1, justified for such kind of sputter deposited smooth films. Finally the interface diffusion parameter ( d ) is obtained from the relation:
where, the interface width defined in Eq. (1) is estimated from the fitting of the specular reflectivity data of the sample. Finally, soft X-ray reflectivity at water-window wavelength of 28 Å has been recorded for Co/Ti multilayer with 60-bilayer periods at the BL-03 soft X-ray beamline at Indus-2 synchrotron source .
3. Results and discussion 3.1 Single layer films: Initially, several single layer Co and Ti films have been deposited in order to optimize different process parameters such as cathode power, Ar flow rate and trolley speed to obtain smooth films with low surface roughness. As has been mentioned above, in the present work Ti and Co layers of nominal thickness of 9 Å have been targeted to achieve a bi-layer thickness of
~18 Å in the multilayers. As described in the previous section, deposition of such low-thickness film is realized by using a mask below the target and thus by reducing the plasma exposure area onto the substrate. As analysis of GIXR spectrum of a film with such ultra-low thickness to obtain its thickness and roughness is not feasible, for process optimization purpose, relatively thicker films of ~100 Å have been deposited by carrying out multiple passes of the trolley below the masked target during deposition, keeping all other deposition parameters fixed. The measured GIXR spectra along with the best-fit theoretical simulations of one representative single layer Ti film and one representative single layer Co film, each deposited with 10 passes of the substrate trolley are shown in fig.3. It has been observed that smooth films of desired thickness and density have been obtained with a trolley speed of 34 mm sec-1, Ar flow rate of 40 ml min.1 and magnetron power of 400 W in case of Ti film and with trolley speed of 51 mm sec1
, magnetron power of 300 W and with a mixed flow of 100 ml min. -1 of Ar and 15 ml min.-1 of
dry air in case of Co films. It should be noted that after carrying out extensive work on magnetron sputter deposited Co/Ti system in our laboratory [5,13,15] we have established that better quality multilayers are grown when Co layers are deposited under a mixed ambience of Ar and dry air than under pure Ar.
3.2 Co/Ti Multilayer films: 3.2.1 Specular X-ray Reflectivity: With the optimized process parameters obtained by characterization of the single-layer films, we have prepared Co/Ti multilayers having nominal bi-layer thicknesses around 18 Å with 10, 20, 40, 60 and 80 number of bi-layers. GIXR data of the samples had been measured in a laboratory reflectometer with the Cu Kα source at 1.54 Å wavelength and the fitting has been
carried out using a genetic algorithm of the GenX program  following Parrat‟s formalism . In order to fit the measured GIXR data for such multilayers, generally theoretical reflectivity plots are generated by modelling with a three-layer structure on Si substrates as mentioned in our earlier study . The first Ti layer on the substrate and top Co layer have been considered two separate layers in the model to take care of the initial instability in deposition and top surface roughness respectively. The second layer has been considered to be a Co/Ti bi-layer repeated (N-1) times for N number of bi-layer whose parameters are kept constant throughout the multilayers. Along with an average interface width
at each interface is introduced in order to
take care of the deviation from ideal abrupt and flat interfaces i.e., the interface imperfections. However, it is observed that this structure is not able to fit our data properly not for even smallest (N=10) number of bilayers. The best-fit theoretical plot generated with this model (represented schematically in fig. 4 (a)) is shown in fig. 5(a). While this model can predict angular positions of the Bragg peaks and Kissing fringes correctly, it fails to match their amplitudes, especially for higher grazing angles of incidence. In order to match the reflectance result across the whole angular range, a CoTi intermixed layer is assumed at Co-on-Ti interfaces, as shown schematically in fig. 4 (b). However, it is seen that this model also fails to fit the data, even the deviation (fig. 5 (b)) from measured data is still higher. Finally, the best fit of the hard X-ray GIXR data has been found when it is assumed that that Co diffusion into Ti layer is not limited within the interface, rather it spreads throughout the entire Ti layer. This leads to effective Ti layer density to increase and effective Co thickness to decrease. This model is schematically shown in fig.4(c) while the corresponding best fit GIXR plot is shown in fig. 5(c). Thus it implies that for such ultra-short Co/Ti multilayer, diffusion of Co into Ti plays the key role in determining the quality of the multilayer. It should be mentioned here that in our earlier
communications also [13,15] on Co/Ti multilayers it has been reported that diffusion at Co-on-Ti interface is higher than that at Ti-on-Co interface. Fig.6 shows the specular GIXR data of Co/Ti multilayer samples having 10, 20, 40, 60 and 80 bi-layers along with the best-fit theoretical plots. We initially tried to fit the GIXR data for all the samples with a three-layer structure with diffusion model as described by fig. 4(c) above assuming that the parameters of the bi-layers remain identical across the depth of the multilayer. However, it has been observed that as the number of bi-layers increases beyond 20, even this model fails to yield any reasonably good fitting which implies that the structure evolution and diffusion properties do not remain same as more and more layers being added in the multilayer structure. This is an indication that we need to vary the interface roughness as well as diffusion properties with increasing number of layers. Such variation in interface imperfection pattern with increasing number of bi-layers have been observed in our earlier study also  for Co/Ti multilayer having 42 Å bi-layer thickness, where cross sectional TEM micrographs clearly depict that, as the growth of the multilayer proceeds and number of bi-layer increases, interface roughness increases and wavy pattern in the growth becomes more and more prominent. There is an accumulation of interface roughness in these Co/Ti multilayers, which propagates across the interface as more number of layers is being added. With this background, in case of the present multilayer system also the interface diffusion is varied as more number of layers being added. The GIXR data of 40-bilayer Co/Ti multilayer is best fitted with a 2-layer model, schematically shown in the inset of fig.7, where total number of bi-layers is divided into two parts whereas for the rest two samples (N=60 and 80), the multilayers are split into 3 parts. The best fit theoretical simulations have also been shown alongwith the respective GIXR data in fig.6. It should be
noted that further division of a multilayer structure into smaller parts for the respective sample does not improve the fit significantly. In order to further express the evolution of interface imperfections of these Co/Ti multilayers quantitatively, we define a density contrast quantity as follows:
i.e., ratio of density difference between Co and Ti of a particular layer in multilayer to density difference between Co and Ti for ideal case. The variation in density contrast of the multilayers with number of layers is plotted in fig. 7, whereas the best fit thicknesses are also given in the inset of the figure. Two important observations can be made from the above results: first it has been observed that as number of layers increases, density contrast at the interface reduces. Consequently, Bragg peak reflectivity decreases with addition of layers for the multilayer with N=80 compared to that with N=60, as can be seen from the inset of fig.6. Further the thickness of Co reduces keeping the bi-layer thickness almost same. This significant variation in growth properties is attributed towards increase in inter-diffusion of Co and Ti layers as the number of bi-layers is increased. It is also clear from the Scattering Length Density (SLD) variation of the multilayers. SLD is a measure of the scattering power of a material which increases with the physical density contrast as well as the intrinsic scattering power of the 'scattering entities' . If the consecutive layers show high electron density contrast, then SLD change will be sharp whereas if there is diffusion of layers into each other leading to low electron density contrast, SLD change will be gradual and smooth. The SLD for the sample with highest number of bilayers (N=80) is plotted (extracted from GenX program) in fig. 8. The SLD variation clearly
demonstrates that density contrast between Co and Ti reduces with addition of more number of layers. It has been found from the above study that diffusion across Co-Ti interface plays a major role in the present multilayer systems. However, another important cause of interface imperfection is interface roughness, which we had seen earlier plays a major role in case of Co/Ti mutli-layer samples with relatively larger bi-layer thickness of 42 Å . As has been mentioned above, the two interface imperfection parameters, viz., interface diffusion d and interface roughness r cannot be separated out by specular X-ray reflectivity measurement. So for complete characterization of the interface between two consecutive layers, all the samples subsequently have been characterized using X-ray reflectivity in non-specular or diffused mode also.
3.2.2 Diffused X-ray Reflectivity: The diffused X-ray reflectivity of the samples has been measured around the 1 st Bragg peak position in the rocking scan geometry. The measured diffused reflectivity data, presented in terms of the quantity
1 dI ( ) as defined above and further normalised with respect to I 0 d 0
respective peak value, are plotted in fig. 9. A qualitative comparison of the diffused spectra can be done from the area under the curve. It is evident from fig. 9 that the area under the curve is maximum for sample with N=10 and as the number of layers increases the area under diffused reflectivity curve steadily decreases. It should be noted that only the physical roughness present at the interfaces of a multilayer contributes towards the diffused scattering and not the interface diffusion. Thus the above observation clearly indicates that physical interface roughness
decreases with increase in number of layers in the present multilayer system implying that there is some sort of smoothening effect with increase in number of bi-layers in these ultra-short period multilayers. This further implies that some of the intensity that is scattered by interface roughness in diffused directions for samples with less bi-layer numbers, should get reflected in the specular direction in case of samples with higher bi-layer numbers and thus Bragg peak reflectivity should increase significantly with increase in N. However, as can be seen from the inset figure of fig.6, Bragg peak reflectivity enhancement with increase of number of layers is not so significant for these ultra-short Co/Ti multilayers. This is an indication that although physical interface roughness decreases with increase in number of layers, contribution from the inter-diffusion between two layers also increases significantly. As inter-diffusion decreases the density and hence refractive index contrast between spacer and absorber layers, it ultimately limits the reflectivity obtained from the multilayer. This observation thus confirms our earlier speculation made from specular reflectivity data that the contribution of interface diffusion ( d ) is higher compared to the interface roughness ( r ) in the total interface imperfection ( ) and plays a dominant role in determining the reflectivity of these ultra-short Co/Ti multilayers. In order to further quantify the interface properties, diffused reflectivity of the whole multilayer structures are computed by the first order DWBA and fitted with measured diffused reflectivity spectra using the two fitting parameters for each interface r and || . d values are subsequently determined from the average values of for the particular sample obtained from fitting of specular reflectivity data and using equation (4) mentioned above. The diffused reflectivity spectrum along with its best fit for one representative sample is shown in fig. 10 and the best-fit values of r , d and || have been presented in Table 1 for both Ti-on-Co and Co15
on-Ti interfaces. It can be found that the interface diffusion is much less compared to the interface roughness for N=10 bi-layer sample. However, as the number of bi-layer increases, for both the Ti-on-Co and Co-on-Ti interfaces diffusion steadily increases, whereas interface roughness decreases. However, the rate of increase of interface diffusion is much rapid for Coon-Ti interface. Thus it is clear that for these ultra-short period Co/Ti multilayers having bi-layer thickness of ~18 Å, diffusion evolution of Co into Ti becomes rapid with number of layers, possibly because Co diffuses into Ti more easily than the other case and this phenomenon plays the dominant role in determining the behaviour of these multilayers. It has also been found from Table 1 that the in-plane correlation length, increases with the increase in number of bi-layers except for the sample with N=80. This again confirms the smoothening effect of high frequency interface roughness. The in-plane correlation length is closely associated with the atomic arrangements and interfacial morphology. As cumulative increase of interface roughness is not occurring for these ultra-low thickness multilayers, this results in the improvement in low spatial frequency roughness in lateral directions when number of bi-layer periods increase. However, it can also be observed that in-plane correlation length ( || ) values are significantly higher in case of the Ti-on-Co layer compared that of Co-on-Ti layer manifesting higher grain sizes of the Co layer. It corroborates with our earlier observations on Co/Ti multilayer also , where it has been observed that Co layer shows peaks in XRD pattern manifesting their polycrystalline nature while there was no peak observed from Ti layer because of their amorphous microstructure.
3.2.3 Soft X-ray reflectivity: Finally, the performances of the multilayers have been tested with soft X-ray radiation in BL-03 soft X-ray beamline at Indus-2 SRS as has been mentioned above. As Bragg peak reflectivity of hard X-ray GIXR data is found to be maximum for the sample having N=60, soft X-ray reflectivity measurement has also been carried out with this sample. The above multilayers are designed as reflecting mirrors above Ti 2p absorption edge near 2.73 nm in the water window regime of the soft X-ray and hence the soft X-ray reflectivity measurements have been carried out at 28 Å. Since the estimated bi-layer thickness is 18 Å, the first Bragg peak position should nominally appear near 42.5°, which agrees well with experimental result as shown in fig.11. The soft X-ray reflectivity data has also been fitted with the theoretical plot generated using GenX software code with the same model used for fitting of hard X-ray GIXR data (by splitting the multilayer in three parts) and the best-fit theoretical plot is also shown in fig. 11 alongwith the experimental data and the best fit results are shown in Table 2. It has been observed that the interface widths of both Co-on-Ti and Ti-on-Co interfaces at the higher wavelength (soft X-ray) as shown in Table 2 are similar to that obtained at the lower wavelength (hard X-ray) as shown in Table 1. In our of earlier communications on Co/Ti multilayers of relatively higher bi-layer thickness of 100-200 Å for neutron supermirror applications , it can be seen that both interface diffusion and interface roughness contribute to the interface imperfections of the multilayer samples and when interface roughness is decreased by introducing dry air during deposition of Co layers, the contact area between the two materials at the Ti-on-Co interface decreases and so inter-diffusion also deceases at the interfaces. In our subsequent communication on Co/Ti multilayers of relatively lower bi-layer thickness of ~42 Å, we had shown that the
interfaces are primarily dominated by interface roughness and a cumulative accumulation of interface roughness takes place with the increase in the number of bi-layers which limits the maximum reflectivity obtained by such kind of multilayers . On the other hand, in the present case of Co/Ti multilayers of ultra-short bi-layer thickness of 18 Å, it has been observed that, the behaviour of the interfaces is significantly different from the previous cases. For such ultra-short bi-layer thickness Co/Ti multilayers, Co diffuse into Ti layers significantly changing the growth morphology drastically. Because of small thickness of each layer (1 nm), diffusion is not limited to interface width, rather it spreads throughout the whole layer. Further, the interface diffusion increases with increase in number of bi-layers and this ultimately results in lower density contrast between the layers and limits the soft X-ray reflectivity obtained from the samples. This information is very important in the context of design and fabrication of water window soft X-ray optics with large number of bi-layers since the reflectivity of the multilayers can be increased by reducing/inhibiting the interface diffusion by using an interface diffusion barrier layer.
4. Conclusion Co/Ti multilayers with 10,20,40,60 and 80 bi-layers having ultra-short (~18 Å) bi-layer thickness, which have applications as reflecting optics in the water window soft X-ray regime of 23-44 Å, have been grown using an indigenously built in-line dc magnetron sputtering system. For experimental realization of deposition of the individual Co and Ti layers having nominal thicknesses as low as ≤ 10 Å, specially designed SS masks have been used below the targets to reduce the exposed lengths of the targets and hence the rates of deposition. Specular and diffused GIXR measurements with hard X-rays have been used to evaluate detailed interfacial structure of
the multilayers. Initially several single layer films have been deposited by varying the deposition conditions and finally with the optimized process parameters obtained by characterisation of the single-layer films, the multilayers were deposited. It has been observed that for the present case of Co/Ti multilayers of ultra-short bi-layer thickness, Co diffuse into Ti layers significantly changing the growth morphology at the interfaces. Because of small thickness of the Ti layers (1 nm), Co diffusion is not limited to interface width, rather it spreads throughout the whole layer. Further, the interface diffusion increases with increase in number of bi-layers and this ultimately results in lower density contrast between the layers. It overshadows the advantage obtained from the reduction in interface roughness with increase in the number of layers and finally limits the soft X-ray reflectivity obtained from the samples. The results are different than that reported by us previously in case of Co/Ti multilayers with high (100 Å) bi-layer thickness or Co/Ti multilayers with moderate (42 Å) bi-layer thicknesses. Whereas is case of the former samples both interface roughness and interface diffusion contribute to the interface imperfections of the multilayer samples, in case the later samples, the interfaces are primarily dominated by interface roughness and a cumulative accumulation of interface roughness takes place with number of periods which limits the maximum reflectivity.
AUTHORS CONTRIBUTION P.Sarkar: Investigation, Formal analysis, Writing Original Draft A. Biswas: Conceptualization, Investigation, Formal analysis S. Ghosh: Investigation S. Rai: Investigation M.H. Modi: Investigation D. Bhattacharyya: Conceptualization, Supervision, Writing-Review & Editing
S. Bajt, J. Alameda, T. Barbee, W. M. Clift, J. Folta, A. Kaufmann, E. A. Spiller, Improved reflectance and stability of Mo/Si multilayers, Opt. Eng. 41 (2202) 1797-1804.
F. A. Harrison, W.W. Craig, F.E. Christensen, C.J. Hailey, W.W. Zhang, S.E. Boggs, D. Stern, W.R. Cook, K. Forster, P. Giommi, et al., The Nuclear Spectroscopic Telescope Array (NuSTAR) High-energy X-Ray Mission, The Astrophys. Journal, 770 (2013) 103.
M. Fernandez-Perea, M. Descalle, R. Soufli, K. P. Ziock, J. Alameda, S. L. Baker, T. J. McCaville, V. Honkimiaki, E. Ziegler, A. C. Jakobsen, F. E.Christensen, M. J. Pivovaroff, Physics of Reflective Optics for the Soft Gamma-Ray Photon Energy Range, Phys. Rev. Lett. 111 (2013) 027404.
 Yurii Uspenskii, Denis Burenkov, Tadashi Hatano, Masaki Yamamoto, Optimal Design of Multilayer Mirrors for Water-Window Microscope Optics, Optical Rev. 14 (2007) 64-73. 
P Sarkar, A. Biswas, R. De, K. D. Rao, S. Ghosh, M. H. Modi, Siju John, H. C. Barshilia, Dibyendu Bhattacharyya, Naba Kishor Sahoo, Performance of Co/Ti multilayers in a water window soft x-ray regime, Appl. Opt. 56 (2017) 7525-7532.
 S. Braun, H. Mai, M. Moss, R. Scholz, A. Leson, Mo/Si Multilayers with Different Barrier Layers for Applications as Extreme Ultraviolet Mirrors, Jpn. J. Appl. Phys. 41 (2002) 4074–4081.  N. Ghafoor, P. O. Å. Persson, J. Birch, F. Eriksson, F. Schäfers, Interface engineered ultrashort period Cr-Ti multilayers as high reflectance mirrors and polarizers for soft X-rays of λ = 2.74 nm wavelength, Appl. Opt., 45 (2006) 137–143.  H. Kumagaia, K.Toyoda, K. Kobayashi, M. Obara, Y. Iimura, Titanium oxide/aluminum oxide multilayer reflectors for „„water-window‟‟ wavelengths, Appl. Phys. Lett. 70 (1997) 2338-2340. 
Q. Huang, Q. Yi, Z. Cao, R. Qi, R A. Loch, P. Jonnard, M. Wu, A. Giglia, W. Li, E. Louis, F. Bijkerk, Z. Zhang, Z Wang, High Reflectance Nanoscale V/Sc Multilayer for Soft X-ray Water Window Region, Sci. Rep. 7 (2017) 12929.
 H. C. Mertins, F. Schafers, H. Grimmer, D. Clemens, P. Boni, M. Horisberger, W/C, W/Ti, Ni/Ti, and Ni/V multilayers for the soft-X-ray range: experimental investigation with synchrotron radiation, Appl. Opt., 37 (1998) 1873-1882.  M. Magnusona, C. F. Haguea, Determination of the refractive index at soft X-ray resonances, J. Electr. Spec. 137 (2004) 519-522.  D. L. Windt, “IMD Version 4.1.1,” 2000, http://cletus.phys.columbia.edu/windt/idl.  A. Biswas, A. Porwal, D. Bhattacharya, C. L. Prajapat, A. Ghosh, M. Nand, C. Nayak, S. Raif, S. N. Jha, M. R. Singh, D. Bhattacharyya, S. Basu, N. K. Sahoo, Effect of dry air on interface smoothening in reactive sputter deposited Co/Ti multilayer, Appl. Surf. Sci. 416 (2017) 168–177.
 J. T. Zhu, S. P. Yue, Y. C. Tu, Y.Z. Zhang, Preparation of Co/Ti multilayer in soft X-ray region by nitrogen reactive sputtering, Opt. Precis. Eng. 23 (2015) 10–14.  A. Biswas, R. Sampathkumar, A. Kumar, D. Bhattacharyya, N. K. Sahoo, K. D. Lagoo, R. D. Veerapur, M. Padmanabhan, R. K. Puri, D. Bhattacharya, S. Singh, S. Basu, Design and development of an in-line sputtering system and process development of thin film multilayer neutron supermirrors, Rev. Sci. Instrum. 85 (2014) 123103.  L. G. Parrat, Surface studies of solids by total reflection of X-rays, Phys. Rev. 95 (1954) 359–369.  D.L. Windt, F.E. Chtirtensen, W.W. Craig, C. Hailey, F.A. Harrison, M. Jimenez-Garate, R. Kalyanaraman, P.H. Mao, Growth, structure, and performance of depth-graded W/Si multilayers for hard x-ray optics, J. Appl. Phys. 88 (2000) 460.  D. K. G. de Boer, X-ray scattering and X-ray fluorescence from materials with rough interfaces”, Phys. Rev. B 53 (1996) 6048.  Mohammed H. Modi, R. K. Gupta, S. R. Kane, V. Prasad, C. K. Garg, P.Yadav, V. K. Raghuvanshi, Amol Singh, Mangalika Sinha, A soft X-ray reflectivity beamline for 1001500 eV energy range at Indus-2 synchrotron radiation source, AIP Conf. Proc. 2054 (2019) 060022.  M. Björck, G. Andersson, GenX: an extensible X-ray reflectivity refinement program utilizing differential evolution, J. Appl. Crystallogr. 40 (2007) 1174–1178.  P. Sarkar, A. Biswas, R. De, K. D. Rao, S. Rai, H. Donthula, D. Bhattacharyya, Effect of ultrathin Cu buffer layer on interfaces of Co/Ti multilayer for use in water-window region, AIP Conf. Proc., 2115 (2019) 030297.
 J. Daillant, A. Gibaud, X-ray and Neutron Reflectivity Principles and Applications, Springer, Berlin Heidelberg, 2009.
Caption of figures:
Schematic diagram of top view of sputtering target with mask.
Schematic diagram showing different configurations of non-specular (diffused) GIXR measurement.
Fig. 3: Specular GIXR data alongwith best-fit theoretical plots of representative single layer films of Co and Ti.
Schematic models of the interfacial morphology of Co/Ti multilayers used for fitting of specular GIXR data: (a) Co/Ti structure without any inter-mixing zone at the interface (b) Co/Ti structure with an intermixing zone at the Co-on-Ti interface and (c) Co/Ti structure assuming Co diffusion across bulk of Ti layer.
Specular GIXR data for the 10-bilayer multilayer sample along with best fit theoretical plots generated using different interfacial morphologies. 25
Specular GIXR data of Co/Ti multilayers with 10, 20, 40, 60 and 80 bi-layers alongwith best fit theoreical plots.
Density contrast variation with number of layers for different multilayer samples as obtained from fitting of specular GIXR data (inset: Schematic model of the Co/Ti multilayer structure used for fitting with best-fit thickness values). 27
Variation of scattering length density with thickness for a Co/Ti multilayer sample with 80 bi-layers.
Non-specular (diffused) X-ray reflectivity data of
different number of bi-layers, around 1st Bragg peak position in rocking scan geometry
Non-specular (diffused) X-ray reflectivity data along with best fit theoretical simulation of a representative Co/Ti multilayer with 80 bilayers around 1st Bragg peak in rocking scan geometry.
Soft X-ray reflectivity data of (Co/Ti) multilayer with 60 number of bi-layers measured at 28 Å wavelength alongwith best-fit theoretical plot.
Table 1: Best fit results of X-ray diffused reflectivity fitting for Co/Ti multilayer with different number of bi-layers.
No. of Bilayers
In-plane correlation Length || (Å)
Interface Roughness r (Å)
Interface Diffusion d (Å)
In-plane correlation Length || (Å)
Interface Roughness r (Å)
Interface Diffusion d (Å)
Table 2: Interface width ( ) obtained from soft X-ray reflectivity fitting of the Co/Ti multilayer with 60 number of bi-layers.
No. of bilayers
for Co-on-Ti interface
for Ti-on-Co interface
(Å) 5.8 Å
(Å) 5.0 Å
Declaration of interests
☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
☐The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: