Microelectronic Engineering 123 (2014) 4–8

Contents lists available at ScienceDirect

Microelectronic Engineering

journal homepage: www.elsevier .com/locate /mee

A scalable anti-sticking layer process via controlled evaporation

http://dx.doi.org/10.1016/j.mee.2014.04.0020167-9317/� 2014 Elsevier B.V. All rights reserved.

⇑ Corresponding author. Tel.: +49 202 439 1629; fax: +49 202 439 1804.E-mail address: c.steinberg@uni-wuppertal.de (C. Steinberg).

Christian Steinberg a,⇑, Khalid Dhima a, Daniel Blenskens a, Andre Mayer a, Si Wang a, Marc Papenheim a,Hella-Christin Scheer a, Joachim Zajadacz b, Klaus Zimmer b

a Microstructure Engineering, Faculty of Electrical, Information and Media Engineering, University of Wuppertal, Wuppertal D-42119, Germanyb Leibniz-Institute for Surface Modification IOM, Leipzig D-04318, Germany

a r t i c l e i n f o

Article history:Received 25 October 2013Received in revised form 11 March 2014Accepted 2 April 2014Available online 13 April 2014

Keywords:Anti-sticking treatmentThermal evaporationUniformityScalability

a b s t r a c t

We developed a novel process for the deposition of anti-sticking layers from the gas phase based on theuse of evaporation cells. The cells are prepared from Teflon and consist of small silane containers with anorifice. Temperature is used to provide a silane-containing gas phase in the containers; the temperature iswell below the boiling point of the silane. The number of cells can be varied to improve the uniformitywith respect to surface energy, to scale-up the process to larger diameters and to reduce the processingtime required. The impact of cell number, processing temperature and processing time is investigatedwith respect to uniformity and with respect to the minimum times and temperatures required. The con-cept works well and shows potential for the control of surface energy beyond anti-sticking purposes.

� 2014 Elsevier B.V. All rights reserved.

1. Introduction

Nanoimprint as a contact process ultimately requires anti-stick-ing layers in order to provide a low adhesion and thus a preferredseparation at the stamp-polymer-interface. It is well known that inorder to be long-living [1], these layers have to be bonded to thestamp surface. Additionally, they should be thin (preferably mono-layers) in order not to affect the dimensions of the stamp and thusthe dimensional fidelity of the replication [2,3]. Fluorinated silanesare suitable materials, as (i) they provide the low surface energies(typically 15–20 mN/m) required for low adhesion, as (ii) they canbe attached to an OH-terminated Si- or SiO2-surface via chemicalbonds and as (iii) they form monolayers in the nanometer rangeas long as self-polymerization of the silane is avoided by providinga water-free environment during processing. Being independentfrom solvent wetting of small capillaries, deposition from the gasphase is superior to a deposition from the liquid phase, in particu-lar with stamp cavities in the nanometer range [2,3].

Gas phase deposition processes are well described and wellcharacterized in the literature. The process according to Beck [2]works at atmospheric pressure, where the silane is droppedbesides the sample in a Petri dish and deposition proceeds at200–250 �C; the process described by Jung [3] operates with a flowof evaporated silane under vacuum, and the whole process consistsof three consecutive steps of silane and water evaporation; a fur-

ther process described by Schift [4] also works under vacuum,where the liquid silane is injected into the vacuum chamber viaa syringe through a membrane. Processes combining vacuum andtemperature are also described [5].

We developed a different process based on the use of evapora-tion cells. The cells are prepared from Teflon and consist of smallsilane containers with an orifice. Temperature is used to providea silane-containing gas phase in the containers, but the tempera-ture chosen is well below the boiling point of the silane. Thus alow, but constant volume stream of silane is provided throughthe orifice throughout the whole process if a sufficient amount ofsilane is deposited in the container at the beginning of the process.This is different from the gas phase process of Beck [2], where thesilane volume provided quickly decreases with deposition time inan uncontrolled way as the silane droplet surface area decreases.Our deposition process is slow and allows the formation of a uni-form layer. This is also different from a situation, where the liquidsilane evaporates in the vacuum abruptly, like in the process ofSchift [4]. In-house experiments have shown that direct evapora-tion into a vacuum, without valve control, results in splatteringand may lead to silane clusters at the sample surface. The vacuumprocess of Jung [3] avoids this issue. There, the silane is evaporatedin a remote cell and the silane vapor is guided to the processingchamber, where a constant stream of silane is provided duringthe deposition time (3 times for 30 min). Our process is similarto the process of Jung, as we want to provide stationary, well-con-trolled deposition conditions throughout the whole processingtime. But the components are simple and do not require corro-

C. Steinberg et al. / Microelectronic Engineering 123 (2014) 4–8 5

sion-resistive vacuum components (during silane binding HCl isformed). In addition, our process requires limited resources ofsilane, only. The whole process is operated within a capped Petridish and does not suffer from corrosion of non-glassy componentsof the deposition system.

In addition, our process is scalable. Scalability here concernstwo issues. First, we intend to develop a simple process that isup-scalable for samples of 10–15 cm diameter and provides uni-form layers. To our knowledge, no uniformity investigation ofanti-sticking layers is presently at hand. Second, we intend todevelop a process that also allows scaling the surface energy itself.This matters for an application beyond the preparation of simpleanti-sticking layers, where low surface energies (in the range of10–20 mN/m) are asked solely. For example, with surface prepara-tion for block copolymer self-assembly, the surface energy has tobe tuned in the range of 25–40 mN/m in order to provide a ‘neutral’non-preferential surface [6]. Thus, scalability of the sample size aswell as of the surface energy is addressed here.

For both purposes, the concept of evaporation cells is testedhere by investigating the impact of processing temperature andof processing time. In view of the uniformity of the layer across10 cm diameter samples also the number of evaporation cells isvaried. The concept is well suitable to provide uniform, low energyanti-sticking layers, working even well with undercut trenches [7].The concept is also well suitable for a scaling of the surface energyin a time-controlled process at adequate temperature.

2. Experimental

As an anti-sticking agent CF3(CF2)5(CH2)2SiCl3 (Fluorooctatri-chlorosilane, FOTS, Sigma–Aldrich) was used. Its boiling point atatmospheric pressure is at 192 �C (data sheet). Literature values[8] (vapor pressure of 0.3 Torr at room temperature) allow to pro-vide curves of the vapor pressure as a function of temperature asshown in Fig. 1, assuming a constant enthalpy of evaporationwithin this temperature range. As our deposition process includesa water step, also the vapor pressure for water is given.

The deposition procedure consists of a UV-exposure (172 nm)to generate OH-groups at the Si surface in an air environment ataverage humidity [9], the thermal evaporation of the silane itself,followed by a thermal evaporation step with DI-water. DuringFOTS evaporation the fluorinated silane chemically bonds to theOH-terminated Si surface and forms a monolayer under the forma-tion of HCl. The subsequent water evaporation provides additionalOH-groups to the silane by reacting with the Cl-groups still present– it has to be assumed that only one of the three Cl-bonds forms abridge to the Si surface. In a final temperature step the OH-groupsreact among themselves to crosslink the layer and to provide long-

Fig. 1. Vapor pressure dependence on temperature for DI-water and FOTS. Thecalculation for FOTS is based on a boiling point of 192 �C at ambient pressure and avapor pressure of 0.3 Torr at room temperature [8].

term stability. All deposition steps are performed in a glove boxunder dry nitrogen flow to provide controlled humidity and thusto avoid a self-polymerization of the FOTS.

The evaporation cells are prepared from Teflon, a non-corrodingmaterial, and consist of small containers with a definite orifice(0.8 mm ø, 2.4 mm long) in a detachable top cover (see Fig. 2a).Typical loading for one cell is about 50 ll of FOTS. Within the con-tainer the initial liquid height is about 1 mm and offers a constantevaporation surface of about 0.5 cm2 throughout the process –with a loading of 50 ll there is still FOTS remaining in the con-tainer after a 90 min process at 120 �C. (The contact angle of FOTSto Teflon is around 35�, thus the container is wetted.) The totalamount of FOTS atoms per cell loading is about 1020 (molar mass480 g/mol, density 1.3 g/cm3). The orifice limits the outflow ofFOTS-vapor from the cell. With the geometries realized (seeFig. 2a) the maximum solid angle for direct transmission amountsto X = 0.4 rad (6% of the full solid angle of 2p). Within the smallcontainer volume (0.45 cm3) the silane vapor pressure is almostinstantly provided, and the small orifice guides only a smallamount of the evaporated FOTS molecules into the processingchamber, the Petri dish, so that the equilibrium within the cellsis only marginally affected. (Similar cells are used for the subse-quent water process. Here, the DI-water loading is uncritical(300 ll) and the cells are used without the cover lid.) The free pathfor N2 collisions at atmospheric pressure is in the range of somemicrometers, only. We expect that collisions of FOTS with N2 areelastic and do not deplete the amount of FOTS available for depo-sition. As all walls have temperatures below the boiling point ofFOTS, adhesion will occur there, with potential re-emission whenno chemical bond is formed. In order to assure constant processingconditions, the orifice is cleaned before every process.

For an up-scaling to samples of 100 mm diameter up to 12 cellsare arranged symmetrically around the wafer, held in place by aring (see Fig. 2b) The whole assembly is operated in a closed Petridish (diameter 16 cm, inner height 3 cm).

The procedure for anti-sticking layer deposition is as follows.100 mm diameter Si substrates are prebaked for 15 min at 200 �Con a hotplate in order to evaporate the H20 from the surface, fol-lowed by UV-exposure for 2 min (XERAFEX�20, OSRAM), 3 mmapart from the lamp housing. For thermal evaporation, FOTS is sup-plied from up to 12 evaporation cells. The whole assembly isplaced in a closed Petri dish and transferred onto a pre-heated hot-plate in a glovebox purged with dry N2. After the processing timeenvisaged the wafer is transferred inside the glovebox into a sim-ilar second Petri dish prepared with the same number of cells, butloaded with DI water. As soon as the hotplate has reached the pro-cessing temperature for the water process (50 �C), the second Petridish is placed on the hotplate and processed for 10 min. The finaltemperature step is performed on a hotplate outside the glovebox,

small orifice

10mm

8mm

Teflon

2.4mm

0.8mm

chamber20m

m

FOTS

Fig. 2a. Schematics of evaporation cell. The evaporation cell is made from Teflon, anon-corroding material. It consists of the container, providing the supply ofvaporized material, and a removable top-cover with a small orifice, providing aconstant outflow of the vapor.

1

evaporation cell

spacerring

100 mm ø wafer

2

3

4 7

5

6

8

9

Petri-dish

measurement points

Fig. 2b. Deposition assembly for 100 mm diameter wafers. A maximum of 12evaporation cells is equally spaced around the wafer within a Petri dish. Foruniformity characterization the wafer was cut into 9 pieces as shown, and 5measurement points were evaluated for each piece as indicated for the case of #5.

16

14

12

10

8

6

4

2

0surf

ace

ener

gy

γ / m

N/m

1 cell 2 cells 3 cells 4 cells 6 cells 12 cells

wafercell

position

1 ... 9

Fig. 3. Uniformity over wafers of 100 mm diameter using a varying number of cellsloaded (process parameters: 120 �C, 90 min). The surface energy was obtained fromcontact angle measurements with DI-water and diiodomethane, the loaded cells arearranged symmetrically, as sketched. The uniformity is quite excellent with morethan 4 cells. Gray bars with 4 cells: see discussion in Section 4.

Fig. 4. Impact of processing temperature and processing time on the surface energyobtained with 4 cells (the error bars refer to 5 different measurements with thecenter piece, #5, see Fig. 2b). The initial surface energy after UV treatment is about66 mN/m. Surface energies less than 20mN/m are suitable to serve as an anti-sticking layer. For a time of 90 min the surface energy is almost constant and as lowas 10–13 mN/m.

6 C. Steinberg et al. / Microelectronic Engineering 123 (2014) 4–8

at 150 �C for 15 min, followed by a temperature ramp to 250 �C(about 5 min).

Characterization of the FOTS monolayer was performed bycontact angle measurement (EasyDrop, KRÜSS) with two differentliquids, DI-water and diiodomethane (CH2I2), to allow thedetermination of the surface energy of the treated sample. TheYoung–Laplace fit of the droplet geometry was used for contactangle determination. The evaluation of the surface energy exploitsthe geometric mean [10,11] and considers the dispersive and polarpart of the respective energies (c = cd + cp). The values used are[12]: water: c = (21.8 + 51) mN/m; CH2I2: c = (48.5 + 2.3) mN/m.Our evaluations show that a fully automated contact angle deter-mination is critical. With contact angles around 90� an incorrectvalue of the contact angle results in large differences in surfaceenergy. Therefore, with contact angles around 90� all automaticevaluation results were inspected and corrected (position ofthree-phase contact line), if necessary.

3. Results

To test the scalability of the approach for large area depositionthe uniformity of the surface energy across a 100 mm wafer wasmeasured. For this purpose the treated wafers were cut in ninepieces (see Fig. 2b), and contact angles were determined in 5 differ-ent positions on each piece. The mean values for each piece aregiven in Fig. 3. The process parameters temperature and time wereheld constant (120 �C, 90 min) at values typical of a successfulpreparation of anti-sticking layers with smaller samples. To inves-tigate the number of cells required for a uniform deposition, thenumber of loaded cells was varied. As Figs. 2a and 2b shows, tworegimes can be identified. With less than 4 cells, the depositionis inhomogeneous. With 4 cells or more the surface energy staysat ± 4% uniformity and values of around 10 mN/m can be achieved,well suitable for anti-sticking armament. Uniform deposition of anefficient anti-sticking layer across 100 mm wafers thus requires atleast 4 loaded cells in our arrangement.

To investigate the temperature behavior of the thermal processthe number of cells was fixed to 4, the minimum number for uni-form deposition. The temperature was varied from 80 �C to 150 �Cwith different processing times (15–90 min). The result is shown inFig. 4. Before deposition the surface energy of the UV-treated Siamounts to 66 mN/m. For a processing time of 15 min the surfaceenergies decrease only slightly with increasing processing temper-ature. To enter a regime adequate for anti-sticking behavior theprocessing time has to be increased. With a long processing time(90 min) the temperature choice is no longer relevant; indepen-

dent of temperature the surface energies obtained are in the rangeof 10–13 mN/m. With intermediate processing times a decreasingeffect of the processing temperature is found. This is in accordancewith the expectations, as the vapor pressure increases in a non-lin-ear way with increasing temperature (see Fig. 1). For anti-stickingpurposes (surface energy less than 20 mN/m) at least 35 min oftreatment time and a processing temperature of 120 �C arerequired.

The results confirm a further interesting application for thethermal deposition process in this configuration beyond the prep-aration of anti-sticking layers. Under adequate processing condi-tions the surface energy itself is scalable to provide optionalvalues in the range of 10–60 mN/m. This is e.g. interesting forthe preparation of substrates for the self-assembly of block copoly-mers, requiring energies in the range of 20–30 mN/m to define a‘neutral’ surface [6]. In particular in the temperature regime of120–150 �C the surface energy can be adjusted by the choice ofthe processing time, only.

To complement the results obtained the time behavior wasstudied. For this study, we chose a temperature of 120 �C, lyingin a range, where temperature changes are of only minor impactto improve the suitability for anti-sticking purposes (see Fig. 5).

Fig. 5. Impact of processing time on the surface energy obtained with 120 �C andthree different cell configurations (the error bars refer to 5 different measurementswith the center piece, #5, see Fig. 2b). With a deposition time less than 15 min nochange of the surface energy is detectable. After a time of about 50 min the surfaceenergy is well below 20 mN/m, adequate for anti-sticking purposes.

C. Steinberg et al. / Microelectronic Engineering 123 (2014) 4–8 7

2, 4 and 12 cells were loaded. Up to about 15 min, there is nostrong decrease in surface energy compared to a UV-treated sub-strate without FOTS deposition. A potential reason for such a timelag could be that a thermal equilibrium is not yet reached in thedeposition chamber. With 12 cells, the time lag appears to beslightly smaller.

For times between 15 min and 35 min a substantial decrease insurface energy occurs. In this time window the surface energydecreases the quicker the more cells are loaded. With 2 cells a dis-tinct step occurs after a process time of 35–40 min. Around thisstep a control of the surface energy in a time-controlled processis hardly feasible. Furthermore, the surface energies reported inFig. 6 for 2 cells may not be reliable due to the limited uniformityobtained with 2 cells only – for Fig. 6 only the central area (#5 inFig. 2b) was evaluated in 5 locations. Taking account of the errordue to non-uniformity the 2-cell results in Fig. 6 suggest a pro-longed time lag compared to the situation with 4 cells.

After about 35 min, the surface energy obtained is 620 mN/m,largely independent of the numbers of cells, and, in addition, lar-gely independent of the deposition time. Thus a processing timeof 40–50 min should be sufficient to provide high quality anti-sticking properties. Admittedly, the deposition may be less uni-form, as a reduced time at temperature is given for the filling of

0 5 10 15 20 25 30 35

FO

TS

ato

ms

sup

plie

d

time / min

1 cell 2 cells 4 cells 6 cells 12 cells

1.1020

8.1019

6.1019

4.1019

2.1019

0

75

65

55

54

35

25

temperature

Fig. 6. Temperature increase with time, measured inside a cell (PT100). With a settemperature of 120 �C of the hotplate a final cell temperature of about 72 �C isreached. Further curves: Estimated total amount of FOTS atoms supplied to the Petridish with different numbers of cells loaded (see text). An increased number of cellsreduce the time required to reach equilibrium conditions.

potential voids – the uniformity investigation was performed witha processing time of 90 min.

4. Discussion

In order to get further insight into the physics behind the timelag, temperature measurements were performed. Fig. 6 shows thetemperature increase with time, when the Petri dish with all con-tainers is placed on the preheated hotplate (120 �C). The measure-ment was performed by means of a PT100 element, located in oneof the containers, with liquid loading. The measurement showsthat about 35 min pass until the final temperature is reached. Thisis due to the thermal load consisting of the Petri dish and the con-tainer-assembly and due to the low heat conductance (0.24 W/km)of the Teflon containers. This time correlates well with the resultsof Fig. 5. The final temperature reached in the cell is about 72 �C.This has to be expected, as we heat from the bottom only – a tem-perature measurement at the Petri dish cover reads�45 �C. The gasphase inside the Petri dish has a similar temperature as the liquidin the containers, about 70 �C.

With this temperature behavior the emission from the cells wasroughly estimated. We assumed the internal container to featurethe vapor pressure corresponding to the actual temperature mea-sured because of the small volume and the even smaller orifice.The outflow from a cell was estimated from the vapor pressureinside the cell and the geometries of the orifice (solid angle, area),as an upper limit. The additional curves in Fig. 6 give the totalamount of FOTS atoms provided to the Petri dish (time integralof the out-flow from the cells at vapor pressure), when differentnumbers of cells are used. With one cell only the correlationbetween the total amount of FOTS atoms supplied and the temper-ature is obvious. An increase of the cell number (see Fig. 6) clearlyaccelerates the process and reduces the time required to establishequilibrium conditions within the Petri dish. Under equilibriumconditions, the partial pressure of FOTS in the Petri dish amountsto the vapor pressure (4 � 10�3 bar at 70 �C), corresponding to5 � 1019 FOTS atoms in the dish in total. This number of FOTSatoms has to be supplied at least from the cells (the estimatednumber in Fig. 6 refers to the supply only, without any loss). Thistime reduction with an increased number of cells is again in accor-dance with the measurements (Fig. 5).

If requested, acceleration of the overall time response can besimply realized by reducing the temperature load of the Tefloncontainers by increasing the container depth. Similarly, a differenttemperature of the sample is easily provided by placing a Teflonslice of adequate thickness between the sample and the Petri dish.Thus, the concept of multiple cells within a small glass container, aPetri dish, provides ample alternatives for the independent controlof the temperatures relevant for evaporation (in the cells) anddeposition (at the sample). We anticipate that this concept hashigh potential for the control of surface energies, even beyondthe preparation of anti-sticking layers.

Finally, chemical integrity of the FOTS is addressed. Best prac-tise is to refill the FOTS-flask with dry N2 after every use to avoidan untimely reaction of the Cl-groups with humidity from air. Suchreactions reduce the binding efficiency of the FOTS resulting in animpairment of the surface energy obtained. Such supplementalinformation is included in Fig. 3. Experiments performed within aperiod of about 6 months are compiled there. To demonstrate thesignificance of FOTS-aging they were carried out without N2 refill.The experiments with 4 and 6 cells (resulting in low surface ener-gies) were performed with an almost new flask. The experimentswith 1–3 cells were done with a filling level of about 60% after2 months, the 12 cells experiment at a filling level of about 30%after 4 months and the repeat experiments with 4 cells with an

8 C. Steinberg et al. / Microelectronic Engineering 123 (2014) 4–8

almost empty flask after 6 months. Aging has no impact on unifor-mity; good uniformity is obtained with at least 4 cells loaded. Butthe absolute value of the surface energy obtained increases sub-stantially with decreasing chemical integrity. For use as an anti-sticking layer, where the only requirement is a low surface energyin the range of 10–20 mN/m, this is not a fundamental issue. Butwhen a certain absolute value of surface energy is intended, a care-ful handling of the FOTS with dry N2 refill of the flask after each useis ultimately required.

5. Summary and conclusion

A novel approach for the thermal deposition of an anti-stickinglayer from a FOTS gas phase based on evaporation cells was inves-tigated. An orifice limits the outflow, thus providing a continuoussupply of FOTS during the whole processing time. The number ofcells used can be varied to scale the process to larger sample sizes,to improve the uniformity and to accelerate the process. With sam-ples of 100 mm diameter 4 cells are at least necessary to provideuniform surface energies. With the actual geometries realized, onlyafter a time lag of 15 min the surface energy becomes reduced. Sur-face energies of 10–20 mN/m (as asked for anti-sticking purposes)are achieved with a temperature as low as 80 �C after 90 min, orwith e.g. a processing time of 40–50 min at an increased tempera-ture of 120 �C. The time behavior could be understood by recordingthe temperature in the cells and by respective estimation of thesupply of FOTS-atoms. The approach has high potential not only

for up-scaling to increased diameters (saturation for more than 4cells) but also for the control of the surface energy value, beyondsimple anti-sticking purposes.

Acknowledgment

Partial funding by Deutsche Forschungsgemeinschaft DFG ishighly acknowledged.

References

[1] R.W. Jaszewski, H. Schift, P. Gröning, G. Margitondo, Microelectron. Eng. 35(1997) 381–384.

[2] M. Beck, M. Graczyk, I. Maximov, E.-L. Sarwe, T.G.I. Ling, M. Keil, L. Montelius,Microelectron. Eng. 61–62 (2002) 441–448.

[3] G.-Y. Jung, Z. Li, W. Wu, Y. Chen, D.L. Olynick, S.-Y. Wang, W.M. Tong, R.S.Williams, Langmuir 21 (2005) 1158–1161.

[4] H. Schift, S. Saxer, S. Park, C. Padeste, U. Pieles, J. Gobrecht, Nanotechnology 16(2005) S171–S175.

[5] R. Kirchner, B. Adolphi, R. Landgraf, J.-W. Fischer, Proc. SPIE 7545 (2010).75450U-1–75450U-9.

[6] C.-C. Liu, C.J. Thode, P.A.R. Delgadillo, G.S.W. Craig, P.F. Nealy, R. Gronheid, J.Vac. Sci. Technol. B 29 (2011) 1017–1023.

[7] S. Möllenbeck, Replikation von hinterschnittenen Strukturen (Ph.D. thesis),University of Wuppertal, Der Andere Verlag, Uelvesbuell, 2012.

[8] T.M. Mayer, M.P. de Broer, N.D. Shinn, P.J. Clews, T.A. Michalske, J. Vac. Sci.Technol. B 18 (2000) 2433–2440.

[9] S. Möllenbeck, N. Bogdanski, A. Mayer, H.-C. Scheer, J. Zajadacz, K. Zimmer,Microelectron. Eng. 87 (2010) 1519–1521.

[10] F.M. Fowkes, J. Chem. Phys. 67 (1963) 2538–2541.[11] D.K. Owen, R.C. Wendt, J. Appl. Pol. Sci. 13 (1969) 1741–1747.[12] S. Wu, Polymer Interface and Adhesion, Marcel Dekker, New York, 1982.