silicon interfaces

Adhesion properties of polymer/silicon interfaces for biological micro-/nanoelectromechanical systems applications Manuel Palacio and Bharat Bhushana兲 Nanotribology Laboratory for Information Storage and MEMS/NEMS (NLIM), The Ohio State University, 201 W 19th Ave., Columbus, Ohio 43210

Nicholas Ferrell and Derek Hansford Department of Biomedical Engineering, The Ohio State University, Columbus, Ohio 43210

共Received 22 April 2006; accepted 26 December 2006; published 3 July 2007兲 Polymers are used in biological micro-nanoelectromechanical systems 共BioMEMS/NEMS兲 applications due to their desirable mechanical properties, biocompatibility, and reduced cost relative to silicon microfabrication processes. Understanding the interfacial properties of the films that are used in BioMEMS/NEMS serves as a useful tool in obtaining higher device yield and greater mechanical reliability. In this study, polystyrene 共PS兲 and glycidyl-ether-bisphenol-A novolac polymer 共SU8兲 on silicon substrates were investigated. SU8 is a commonly used material in MEMS/ NEMS fabrication, while PS is evaluated for its potential use in BioMEMS/NEMS for interaction with biological cells. The aim is to examine the delamination of the interfaces. Nanoindentation was employed on the PS/Si and SU8/Si film systems coated with a thin metallic layer of Cr to facilitate delamination. The interfacial adhesion energy was determined from measuring the size of the resulting delamination and the contact radius. Scale effects were investigated by comparing the behavior of thin and thick PS and SU8 films, where a thickness dependence on the interfacial adhesion energy was observed. In addition to room temperature testing, film delamination experiments were conducted at 50 and 70 ° C by fitting the nanoindenter with a heating stage in order to study temperature effects. Nanoindentation-induced delamination is demonstrated for microstrips of PS and SU8 and the measured interfacial adhesion energy is compared to those obtained from films. © 2007 American Vacuum Society. 关DOI: 10.1116/1.2435390兴

I. INTRODUCTION Biological micro-/nanoelectromechanical systems 共BioMEMS/NEMS兲 applications have expanded into a variety of areas such as drug delivery, microfluidics, and chemical and biological sensing.1–8 Various structures have been created either by traditional silicon-based batch processes or the newer polymer-based fabrication techniques.9,10 Polymers are advantageous for certain device applications because they offer a number of advantages relative to silicon. In addition to being relatively inexpensive, many polymers are also biocompatible, so they may be integrated into biomedical devices with minimal detrimental effects to the host or on the biofluids. An improvement in device functionality 共relative to silicon兲 is also possible due to the mechanical properties of the polymer used, which have properties much closer to biological tissues. An example is shown in Fig. 1, which is a polystyrene 共PS兲-based sensor for measuring the mechanical forces exerted by a cell. The adhesive and contractile forces generated by the cell cause the cantilevers in the sensor to deflect which is detected by an optical method. For this application, large, detectable deflections can be observed by selecting a polymer over silicon, thus providing sensitivity to very low magnitude forces.11

Two polymer materials were investigated in this study, PS and glycidyl-ether-bisphenol-A novolak 共referred to by its trade name SU8兲. The chemical structure and physical properties of these materials are listed in Table I. Polystyrene is particularly desirable for BioMEMS due to its ubiquitous use in tissue culture applications. Plasma treated polystyrene is the most commonly used material for in vitro cell biology studies of adherent cells. Thus, there is a wealth of knowledge about cellular behavior on polystyrene. In addition to tissue culture polystyrene, various surface modification techniques can be employed for functionalizing the PS surface in order to promote cell attachment and proliferation.12 Oxygen plasma modified polystyrene has been shown to improve cell growth, proliferation, and expression of cellular adhesion proteins proportional to the surface oxygen concentration.13

a兲

Author to whom correspondence should be addressed; electronic mail: [email protected]

1275

J. Vac. Sci. Technol. A 25„4…, Jul/Aug 2007

FIG. 1. SEM image of a polystyrene-based cell force sensor.

0734-2101/2007/25„4…/1275/10/$23.00

©2007 American Vacuum Society

1275

1276

Palacio et al.: Adhesion properties of polymer/silicon interfaces

1276

TABLE I. Summary of the physical properties of PS and SU8.

a

Reference 36. Reference 37. c Reference 38. d Reference 39. b

The previous knowledge of cellular interactions with polystyrene makes it a logical choice to use in BioMEMS/NEMS devices for cellular interactions. SU8 is a photosensitive material commonly used in MEMS processing. It is employed as a structural material in fabrication of MEMS components.14,15 Previous studies have shown that SU8 exhibits relatively good biocompatibility, making it a strong candidate for biological applications.16 SU8 has been employed in a number of BioMEMS devices.17–19 An example of an SU8 application is shown in Fig. 2. The micrograph shows an image of an SU8 microgripper used for manipulation of single cells in solution. The use of SU8 as the structural material allows the gripper to be electrothermally actuated at low voltage, thus providing an attractive actuator mechanism for operation in ionic solutions such as saline and cell culture medium. The integrity of the interface between the structural components of MEMS devices and the underlying substrate is an important technological issue. Numerous studies are available for the adhesion characterization of metals and inorganic materials due to the demands of various industries such as aerospace and microelectronics. Examples include peel testing, double cantilever beam testing, and four-point bending.20–23 The latter two methods have been developed for macroscopic sandwich structures, where two substrates 共one containing the film of interest兲 are bonded together prior

FIG. 2. SEM image of an SU8-based microgripper 共from Ref. 18兲. J. Vac. Sci. Technol. A, Vol. 25, No. 4, Jul/Aug 2007

to adhesion assessment. From these techniques, either the strain energy release rate 共G兲共referred to in this article as the interfacial adhesion energy兲 or the stress intensity 共K兲 is reported. Indentation can also be employed for thin film adhesion tests. In this technique, the applied load from the indenter initiates and propagates a delamination, the extent of which is related to the measured loads and displacements. An advantage of this method is the ease of sample preparation, relative to the traditional sandwich specimens. The induced fracture allows quantitative assessment of adhesion.24 However, it is difficult to induce delamination if the material is ductile or if the film is strongly adhering. To overcome this limitation, an overlayer is deposited over the film of interest. The overlayer is a material with a high intrinsic stress 共such as Cr兲 which provides additional driving force for delamination.25 This procedure has been successfully employed for spontaneous delamination of metallic thin films and microstrip structures,25,26 as well as for indentationinduced delamination of films.24 There is a paucity of adhesion studies on materials and structures that are relevant for BioMEMS/NEMS applications. Understanding the interfacial properties of the films that are used in BioMEMS/NEMS serves as a useful tool in obtaining higher device yield and greater mechanical reliability. In future device generations, features with dimensions on the order of 100 nm or less will be integrated, so it is important to understand how length scale affects interfacial adhesion. It is important to understand the adhesion properties of three-dimensional structures such as the microstrips since potential BioMEMS/NEMS applications contain thin features on a substrate. The appropriate materials selection, design, and processing changes can be envisaged if the strength of the component interfaces is quantified. This article presents the adhesion characterization between silicon and two polymers that are used in BioMEMS/ NEMS applications, PS, and SU8. The polymer SU8 was evaluated due to its common application as a structural material in MEMS/NEMS and PS was evaluated as a nontraditional material for use in BioMEMS/NEMS for cellular interactions. A Cr overlayer on the polymer films was employed to facilitate the nanoindentation-induced delami-

1277


nation process. Atomic force microscopy 共AFM兲 was used to characterize the size of the resulting blisters, which was then used to calculate the interfacial adhesion energy 共G兲共or the amount of energy available to form two new surfaces as the crack expands兲. The delamination response of these films was examined at elevated temperatures by fitting a heating assembly into the indenter’s sample stage. Ultimately, the goal is to apply this delamination method to polymer microstrips, which represents structures that can be incorporated in a BioMEMS/NEMS device. Indentation-induced delamination of PS and SU8 microstrips on silicon substrates is demonstrated for the first time.

II. EXPERIMENTAL DETAILS A. Materials and sample preparation

Four types of samples were prepared for evaluation, namely, films of PS and SU8 and microstrips of PS and SU8, all on silicon substrates. For PS, the films investigated were 7100, 870, 300, and 120 nm thick. The SU8 films studied were 7400, 1470, 440, and 220 nm thick. For all of the SU8 films and the 7100 nm PS film, a razor blade was used to carefully remove a portion of the film and a stylus profiler 共Veeco, Inc.兲 was used to measure the step height. For the other PS films, a Nanometrics Nanospec 3000PHX film thickness measurement system was used to determine the film thickness. Profiler data are the average of three measurements and Nanospec data are the average of five measurements. This thickness range was selected to investigate the effect of the length scale on the adhesion characteristics of these polymers. Studies on films serve as the benchmark for establishing the appropriate sample preparation and testing procedure for the microstrips. The latter are of interest since these structures can be found in a potential BioMEMS/ NEMS device 共such as the cell force sensor shown in Fig. 1兲. For PS, the microstrips were 3400 and 1300 nm thick, while for SU8, they were 3500 and 2400 nm thick, which are within the thickness range that can be employed in device structures. The thickness of the microstrips was measured using a stylus profiler. To prepare PS films on silicon substrate, PS pellets 共melt flow index of 4.0, Sigma-Aldrich兲 were dissolved in anisole 共Acros Organics兲 to prepare three solutions of varying concentrations 共20%, 10%, and 3% wt/wt兲 in order to achieve the desired film thickness. Silicon substrates with dimensions of 20⫻ 20⫻ 0.5 mm3 were cut from precleaned silicon wafers. Films with four different thicknesses were prepared by spin coating the solutions on the silicon substrates. The 20% wt/wt solution was spin coated at 1000 rpm to prepare 7100 nm thick film, the 10% wt/wt solution was spin coated at 3000 rpm for the 870 nm thick film and the 3% wt/wt solution was used for the 300 and 120 nm films 共at 1000 and 3500 rpm, respectively兲. After spin coating, the wafers were baked at 70 ° C for 2 min to evaporate residual solvent from the films. SU8 2005 共MicroChem Corp.兲 was diluted with SU8 2000 thinner 共MicroChem Corp.兲 for preparation of SU8 JVST A - Vacuum, Surfaces, and Films

1277

films of various thicknesses. Pure SU8 2005 spin coated at 1000 rpm was used to create a 7400 nm film. Solutions of SU8 2005:SU8 thinner with ratios of 3:1, 1:2, and 1:3 spin coated at 3000 rpm were used to create 1470, 440, and 220 nm thick films, respectively. The coated films were exposed to UV and baked to promote cross-linking of the SU8. The SU8 films were prebaked, exposed to UV, and postexposure baked according to the parameters provided in the manufacturer’s data sheets. SU8 microstrips were fabricated using standard photolithography techniques. Photoresist was spin coated on p-type 具100典 silicon wafers. Films were baked to remove the residual solvent and exposed to UV through a chrome/glass photomask with 5 ␮m features to create the microstrips on the exposed areas. These were then processed according to the manufacturer’s suggested parameters. After processing, feature heights were measured using a profiler. Polystyrene microstrips were fabricated using the softlithography-based procedure shown in Fig. 3. Standard photolithography cannot be used because PS, unlike conventional photoresists, is not photosensitive. In addition, PS is thermoplastic such that micromolding is a suitable processing technique. A poly共dimethylsiloxane兲共PDMS兲 mold was first fabricated from an SU8 photoresist master as described previously.27 The SU8 master consisted of 5 ␮m wide lines separated by 45 ␮m. Two different mold depths were used in this work. One mold had 2.3 ␮m deep channels and the other had 5.3 ␮m deep channels. A 10:1 ratio of T-2 PDMS translucent base and curing agent 共Dow Corning兲 were mixed thoroughly and poured over the photoresist master to transfer the pattern into the PDMS. The mold was allowed to cure for 48 h before removal from the master. Solutions of PS in anisole were then spin coated on the PDMS molds. PS solution concentrations ranged from 3% to 7.5% and spin speeds ranged from 1000 to 4000 rpm. After coating, the molds were brought into contact with a glass slide heated to 200 ° C to remove the PS from the raised portions of the PDMS mold. Slight pressure of 17 kPa 共2.5 psi兲 was applied to the top of the mold during the first stamping. To remove the PS from the recessed channels, the mold was applied to a silicon substrate heated to 125 ° C. For the 5.3 ␮m deep mold, pressures ranging from approximately 550– 1100 kPa 共80– 160 psi兲 were used to remove the polymer from the mold. For the 2.3 ␮m deep mold, pressures ranging from 280 to 350 kPa 共40– 50 psi兲 were used. For the 5.3 and 2.3 ␮m deep molds, feature heights ranged from 2.3 to 4.3 and 1.3 to 1.7 ␮m, respectively, depending on the solution concentration. Feature heights were measured using a profiler. All the film and microstrip samples were coated with an overlayer of 600 nm thick Cr using an electron beam 共ebeam兲 evaporator 共Denton Vacuum兲. B. Nanomechanical characterization

The hardness and elastic modulus of PS and SU8 films were evaluated using a Nano Indenter II® 共MTS Systems Corp.兲 in the continuous stiffness mode 共CSM兲 equipped

1278


1278

FIG. 3. Schematic of the polymer microstrip fabrication process: 共i兲 PDMS mold, 共ii兲 PDMS mold after spin coating with polymer 共PS兲, 共iii兲 mold applied to heated glass slide to remove excess polymer, 共iv兲 excess polymer transferred to glass slide, 共v兲 PDMS mold after removal of excess polymer, 共vi兲 PDMS mold is applied to silicon substrate with heat and pressure, and 共vii兲 microstrips are transferred to the silicon substrate.

with a diamond Berkovich tip. In this mode, the dynamic load and displacement of the indenter are monitored during indentation with a force resolution of about 75 nN and displacement resolution of about 0.1 nm. The CSM technique has been described in detail in Li and Bhushan.28 Briefly, a harmonic force, F = F0e j␻t, is added to the nominally increasing load P on the indenter. The displacement response of the indenter at the excitation frequency and the phase angle between the two, h = h0ei共␻t+␾兲 are measured continuously as a function of depth. Solving for the in-phase and out-of-phase portions of the response results in an explicit determination of the contact stiffness S as a continuous function of contact depth, which is the actual depth that the sample is in contact with the indenter tip. The contact depth is obtained by subtracting the elastic displacement of the surface at the contact perimeter from the indentation displacement. In this article, the maximum indentation displacement was controlled to 500 nm and the peak-to-peak load amplitude was 1.2 ␮N at a frequency of 45 Hz. C. Adhesion testing

Four PS 共thicknesses of 7100, 870, 300, and 120 nm兲 and four SU8 films 共thicknesses of 7400, 1470, 440, and 220 nm兲 were indented at room temperature. The delamination was induced using a conical diamond tip with 90° included angle. The range of applied load was 100– 600 mN. The 870 nm thick PS film and the 1470 nm thick SU8 film were also indented at 50 and 70 ° C using the same load used in the room temperature test. In order to heat the sample, the wafer with the film was mounted on a heating stage.29 The heat-generating elements of the heating stage were Ohmic resistors encapsulated in a steel holder. J-type thermocouples were used to measure the sample temperature. J. Vac. Sci. Technol. A, Vol. 25, No. 4, Jul/Aug 2007

A thermal controller and a solid-state relay were used to control the temperature by adjusting on/off time. The indentation-induced deformation of the films was characterized using a NanoScope III 共Veeco, Inc.兲 AFM. The AFM was operated in contact mode. From the section analysis feature of the NanoScope III software, the dimensions of the contact, plastic zone, and delamination radii were obtained. Two PS and two SU8 microstrip samples were indented at room temperature. The PS samples were 3400 and 1300 nm thick, while the SU8 samples were 3500 and 2400 nm thick. The applied load range was from 100 to 200 mN. III. RESULTS AND DISCUSSION A. Nanomechanical characterization

Hardness 共H兲 and elastic modulus 共E兲 plots as a function of contact depth are presented in Fig. 4. Three indents were made on the PS and SU8 films, and the representative data are shown. These measurements are necessary to calculate the interfacial adhesion energy 共G兲 as described later. The H and E values at the near surface are slightly elevated primarily due to error in the calculation of the contact depth at small indentation sizes. This error is based on the assumption that the materials behave as a rigid plastics as opposed to the viscoeleastic nature of the materials used in this study.28 Surface roughness of the sample can also lead to errors in the contact depth for shallow indents. A plateau is reached 共approximately at a contact depth of 200 nm兲 as the indenter penetrates deeper into the material. For subsequent calculations, these plateau values were used as the data at the corresponding depth range are more reproducible. Data at 22, 50, and 70 ° C are shown since the adhesion characteristics at

1279


1279

stress ␴0 at the inner diameter of the delamination, which is the outer edge of the contact 共r = a兲. The stress distributions in the delaminated portion of the film are

␴r = ␴0关1 + ␣c2/r2兴/关1 + ␣c2/a2兴, ␴␪ = ␴0关1 − ␣c2/r2兴/关1 + ␣c2/a2兴,

共1兲

where c is the delamination radius and a is the contact radius 共both obtained from the AFM images兲, r is the radial position, the subscripts r and ␪ refer to the radial and circumferential directions, ␣ is 共1 − ␯ f 兲 / 共1 + ␯ f 兲, and ␯ f is Poisson’s ratio of the film. The radial stress ␴0 is obtained by applying the Tresca yield criteria to the plastically deformed contact zone. The vertical stress in the contact zone is equal to the mean indenter contact pressure, taken to be the film hardness H. From the Tresca yield criteria, the radial stress ␴0 is then equal to 共␴ys,f -H f 兲, the difference between the yield strength of the film 共␴ys,f 兲 and the hardness 共H f 兲 of the film.31 The strain energy U is obtained by evaluating the following integral over the film volume: FIG. 4. Hardness and elastic modulus of 共a兲 7100 nm thick PS and 共b兲 7410 nm thick SU8 films as a function of contact depth. Indents were performed at 22, 50, and 70 ° C.

higher temperatures were also examined. For both PS and SU8, H and E only dropped slightly 共⬍5 % 兲 at 70 ° C. No significant decrease was observed since the test temperatures are well below the glass transition temperature for each polymer.

U=

Indents were made at three different locations on the same film sample. The load-displacement plots, postindent AFM images, and the height profile across the center of the indents for one representative indent for the PS and SU8 films are presented in Fig. 5. The load-displacement curves are for the lowest possible loads where an abrupt change in displacement at constant load is observed. This is generally taken as the onset of film delamination. At loads lower than the excursion point, only a small amount of hysteresis is present and blisters are not seen, indicating that delamination did not occur. Comparing PS to SU8, it is observed that higher loads were necessary to observe delamination in the latter. With regards to film thickness, a substrate interaction effect is observed for both PS and SU8 films thinner than 500 nm. These films appear stiffer since the deformation response is both from the film and the stiffer silicon substrate. As a result, a higher load is necessary to induce delamination in the thinner films. The method for evaluating the interfacial adhesion energy 共G兲 of the film is based on the model of Thouless30 where the debonded portion of the film is modeled as an elastic disk with plane stress conditions in the axial direction. The stress distribution in the disk is determined by applying elasticity solutions with a zero-displacement boundary condition at the outer diameter of the delamination 共r = c兲 and a fixed radial JVST A - Vacuum, Surfaces, and Films

冕

c

a

关␴r2 + ␴␪2 − 2␯ f ␴r␴␪兴rdr,

共2兲

where t f is the film thickness and E f is the film’s elastic modulus. The interfacial adhesion energy G is found by differentiating the strain energy with respect to crack area30 and is given as G=

B. Adhesion testing of films

␲t f Ef

2共1 − ␯2f 兲共␴ys,f − H f 兲2t f . E f 共1 + ␯ f + 共c/a兲2共1 − ␯ f 兲兲2

共3兲

One approach to calculate yield strength of the film is by the imaging method, where ␴ys,f is calculated from the pileup radius b of an indent with a maximum imposed load Pmax as follows:32

␴ys =

3Pmax . 2␲b2

共4兲

Since the in situ determination of the pileup radius is not possible with the indenter system used in this study, it is necessary to use another approach. In another approach used here, the yield strength of PS and SU8 was approximated based on the known relationship between ␴ys and the material hardness.33 The yield strength is related to the hardness by a material-specific constant k such that ␴ys = H / k. The TABLE II. Summary of measured interfacial adhesion energies 共G兲 as a function of film thickness for PS and SU8 at room temperature. PS

SU8

Film thickness 共nm兲

G 共J / m2兲

Film thickness 共nm兲

G 共J / m2兲

7100 870 300 120

0.76 0.21 0.09 0.02

7400 1470 440 220

1.70 0.20 0.08 0.06

1280


1280

FIG. 5. Load-displacement plots, AFM images, and height profile across the center of indents into 共a兲 7100, 870, 300, and 120 nm thick PS and 共b兲 7400, 1470, 440, and 220 nm thick SU8 films.

constant k depends on the state of strain underneath the indenter. For various glassy polymers, it has been found that k ranges from 2 to 3. In this study, the k for PS and SU8 is taken as 2.5. J. Vac. Sci. Technol. A, Vol. 25, No. 4, Jul/Aug 2007

Table II summarizes the room temperature interfacial adhesion energies of PS and SU8 as a function of film thickness by using measured values of H f , E f , c and a, and reported value of v f in Table I. The observed value of G for

1281


1281

FIG. 6. Load-displacement plots, AFM images, and height profile across the center of indents into the 共a兲 870 nm thick PS and 共b兲 1470 nm thick SU8 films taken at 22, 50, and 70 ° C.

PS/silicon is close to that reported for PS/glass 共0.6 J / m2 for a 3.5 ␮m film兲 using Eq. 共3兲.34 For both polymers, thicker films had higher G values, indicating a length scale effect. This is because the interfacial adhesion energy 共or the strain energy release rate兲 has two component mechanisms. First is JVST A - Vacuum, Surfaces, and Films

the work of adhesion, which is the energy of bond separation across the interface. Second is the plastic energy dissipation within the polymer. For the thicker films, the plastic zone increases and this contributes to a higher measured value for the interfacial adhesion energy.

1282


1282

TABLE III. Summary of measured interfacial adhesion energies 共G兲 as a function of temperature for PS and SU8 films. G 共J / m2兲 Temperature 共°C兲

870 nm PS

1470 nm SU8

22 50 70

0.21 0.23 0.32

0.20 0.18 0.22

Figure 6 presents the load-displacement curves, AFM images, and the corresponding section profiles for indents on PS 共870 nm thick兲 and SU8 共1470 nm兲 at 22 ° C 共room temperature兲, 50 ° C, and 70 ° C. As summarized in Table III, the calculated G for the PS films is similar at room temperature and 50 ° C, and then exhibits a slight increase at 70 ° C. For SU8, the G is practically the same at all test temperatures. This observed elevation of the interfacial adhesion energy of polystyrene at higher test temperatures can be accounted for by the increased plastic energy dissipation as polymer chain motion is more likely at higher temperatures. However, this is not applicable to SU8, since it is a cross-linked polymer and chain motion is suppressed. In addition, SU8 has a much higher glass transition temperature 共Table I兲. These reasons account for the insensitivity of the G value of SU8 to these test temperatures. Figure 7 shows the selected high magnification scanning electron microscopy 共SEM兲 micrographs of 120 nm thick PS and 220 nm thick SU8 films after indentation. A crater is

FIG. 8. Schematic of the stages of the indentation-induced delamination for a film-substrate system 共modified from Ref. 35兲.

formed on the material that was in contact with the indenter tip. Around this is the blister which appears to be raised relative to the rest of the film. Along with the AFM images 共Fig. 5兲, this indicates that both the PS and SU8 delaminated in a three-stage mechanism similar to that proposed by Li and Bhushan35 to account for the delamination in diamondlike carbon coatings on silicon, as shown in Fig. 8. In the figure, t is the film thickness, Vi is the volume displaced by the indenter tip, and ␴i is the stress induced by the indentation, which is counteracted by the residual stress ␴r. In this mechanism, the first stage is the compression of material induced by the application of the load, followed by delamination and buckling of the material around the contact area. The third stage is the formation of a ringlike crack on the delaminated film. Radial cracking is observed on the surface. This is attributed to the high applied loads that are necessary to induce delamination. From these images alone, it is difficult to assess whether the radial cracks run through the thickness of the polymer film, or if it is only on the Cr layer. C. Adhesion testing of microstrips

FIG. 7. SEM image of delamination for 120 nm thick PS and 220 nm thick SU8 films. J. Vac. Sci. Technol. A, Vol. 25, No. 4, Jul/Aug 2007

Ultimately, the goal is to characterize microstrips of PS and SU8 since these structures can delaminate from the substrate of a potential BioMEMS/NEMS device due to various processing or service conditions. Figure 9 presents the loaddisplacement curves and SEM images of the PS and SU8 microstrips taken after indentation, where failure of the polymer/silicon interface was observed in all cases. The loading portion of the load-displacement plots starts with a low stiffness 共i.e., low slope兲 that gradually increases as the thickness of the microstrip is approached. This corresponds to the gradual compression of the area being indented in the microstrip prior to failure of the polymer/silicon interface. From the load-displacement plots, it is observed that a lower load is sufficient to induce observable deformation in the thinner PS microstrip 共1300 nm thickness兲 compared to the rest of the sample set. This can be correlated to the extent of indenter penetration into the thickness of the microstrip. The other samples were thicker and required a higher imposed

1283


1283

FIG. 9. Load-displacement plots and SEM image of PS and SU8 microstrips.

load to enable compression of the material. Unlike the thin film samples, a substrate interaction effect was not observed. In the SEM images, the observed cracks at the microstrip/ substrate interface originating from the edge of the indent indicate delamination. Certain areas of the microstrips were

JVST A - Vacuum, Surfaces, and Films

raised, indicating that buckling also occurred. These images demonstrate that indentation can be successfully employed to induce interfacial failure in these polymer microstrip structures. In addition, the images also imply that these microstrips fail in a manner similar to films, as the failed areas of

1284


TABLE IV. Summary of measured interfacial adhesion energies 共G兲 as a function of thickness for PS and SU8 microstrips. PS

SU8

Thickness 共nm兲

G 共J / m2兲

Thickness 共nm兲

G 共J / m2兲

3400 1300

0.53 0.42

3500 2400

0.11 0.08

the microstrips are similar to stage 2 in the schematic shown in Fig. 8 共which was originally proposed to account for film delamination兲. Table IV lists the calculated G values for these four microstrips. The values were determined, using Eq. 共3兲, using the SEM images to determine the contact and delamination radii 共a and c, respectively兲. It is noted that these G’s should be considered as estimates since the model for Eq. 共3兲 is for the evaluation of film adhesion. The microstrip G values are expected to be lower compared to that obtained from the films. It is expected that these strips would be easier to delaminate since a smaller volume of material needs to be displaced. This explains the results for the SU8 microstrips relative to the thin films. Compared to SU8, G values are higher for the PS microstrips. The increased adhesion strength in PS is attributed to the applied heat and pressure in the microstrip stamping process that could lead to a stronger polystyrene/silicon interface. IV. CONCLUSIONS Delamination of PS and SU8 films on silicon substrates has been induced through indentation and the interfacial adhesion energy 共G兲 has been quantified. For both materials, thicker films had higher interfacial adhesion energies due to plastic dissipation in the polymer. The same phenomenon is proposed to account for the observed slight increase in the adhesion energy at elevated temperatures in the non-crosslinked polymer 共PS兲. The SU8 film is not sensitive to the test temperatures used because it has a higher glass transition temperature and is cross-linked. This delamination method, which was originally developed for films, can be applied to quantify the adhesion properties of polymer microstrip structures. Indentation-induced failure of microstrips of PS and SU8 has been demonstrated and characterization of the failed areas reveals that these structures fail in a manner similar to thin films. The higher G of the PS microstrips relative to SU8 is attributed to the difference in the fabrication method used. ACKNOWLEDGMENT The financial support for this project was provided by the industrial membership of the Nanotribology Laboratory for Information Storage and MEMS/NEMS 共NLIM兲. 1

Controlled Drug Delivery: Challenges and Strategies, edited by K. Park 共American Chemical Society, Washington, D. C., 1997兲. 2 Microsystem Technology in Chemistry and Life Sciences, Topics in Cur-

J. Vac. Sci. Technol. A, Vol. 25, No. 4, Jul/Aug 2007

1284

rent Chemistry, edited by A. Manz and H. Becker 共Springer, Heidelberg, Germany, 1998兲. 3 Biochip Technology, edited by J. Cheng and L. J. Kricka 共Harwood Academic, Philadelphia, 2001兲. 4 Integrated Microfabricated Devices, edited by M. J. Heller and A. Guttman 共Dekker, New York, 2001兲. 5 Frontiers in Human Genetics, edited by L. P. San and E. P. H. Yap 共World Scientific, Singapore, 2001兲. 6 Lab-On-A-Chip: Chemistry in Miniaturized Synthesis and Analysis Systems, edited by A. van der Berg 共Elsevier, Amsterdam, 2003兲. 7 P. A. Oeberg, T. Togawa, and F. A. Spelman, Sensors in Medicine and Health Care 共Wiley, New York, 2004兲. 8 B. Bhushan, Springer Handbook of Nanotechnology, 2nd ed. 共Springer, Heidelberg, Germany, 2007兲. 9 M. J. Madou, Fundamentals of Microfabrication 共CRC, Boca Raton, FL, 1997兲. 10 Y. Xia and G. M. Whitesides, Annu. Rev. Mater. Sci. 28, 153 共1998兲. 11 N. Ferrell, J. Woodard, and D. J. Hansford, in Proceedings of the Second IASTED International Conference on Biomechanics, edited by M. H. Hamza 共ACTA Press, Calgary, Canada, 2004兲, pp. 146–149. 12 I. H. Loh, M. S. Sheu, and A. B. Fishcer, in Reference of Functional Polymers: Synthesis and Applications, edited by R. Arshady 共American Chemical Society, Washington, D. C., 1997兲. 13 T. G. van Kooten, H. T. Spijker, and H. H. Busscher, Biomaterials 25, 1735 共2004兲. 14 G. Genolet, J. Brugger, M. Despont, U. Drechsler, P. Vettiger, and N. F. de Rooij, Rev. Sci. Instrum. 70, 2398 共1999兲. 15 L. Gammelgaard, P. A. Rasmussen, M. Calleja, P. Vettiger, and A. Boisen, Appl. Phys. Lett. 88, 113508 共2006兲. 16 G. Voskerician, M. S. Shive, R. S. Shawgo, H. von Recum, J. M. Anderson, M. J. Cima, and R. Langer, Biomaterials 24, 1959 共2003兲. 17 E. W. H. Jager, C. Immerstrand, K. H. Peterson, K. Magnusson, I. Lundström, and O. Inganäs, Biomed. Microdevices 4, 177 共2002兲. 18 N. Chronis and L. P. Lee, J. Microelectromech. Syst. 14, 857 共2005兲. 19 J. H. T. Ransley, M. Wateri, D. Sukumaran, R. A. McKendry, and A. A. Seshia, Microelectron. Eng. 83, 1621 共2006兲. 20 S. Yurenka, J. Appl. Polym. Sci. 6, 136 共1962兲. 21 B. Bhushan, in Testing of Metallic and Inorganic Coatings, Special Publication No. STP 947, edited by W. B. Harding and G. A. DiBari 共ASTM, Philadelphia, 1987兲, pp. 289–309. 22 B. Bhushan and B. K. Gupta, Handbook of Tribology: Materials, Coatings, and Surface Treatments 共McGraw-Hill, New York, 1991兲; reprint with corrections 共Krieger, Malabar, FL, 1997兲. 23 A. A. Volinsky, N. R. Moody, and W. W. Gerberich, Acta Mater. 50, 441 共2002兲. 24 M. D. Kriese, W. W. Gerberich, and N. R. Moody, J. Mater. Res. 14, 3007 共1999兲. 25 A. Bagchi, G. E. Lucas, Z. Suo, and A. G. Evans, J. Mater. Res. 9, 1734 共1994兲. 26 G. Xu, M. Y. He, and D. R. Clarke, Acta Mater. 47, 4131 共1999兲. 27 J. C. McDonald, D. C. Duffy, J. R. Anderson, D. T. Chiu, H. Wu, O. J. A. Schueller, and G. M. Whitesides, Electrophoresis 21, 27 共2000兲. 28 X. Li and B. Bhushan, Mater. Charact. 48, 11 共2002兲. 29 H. Liu and B. Bhushan, Ultramicroscopy 91, 185 共2002兲. 30 M. D. Thouless, Acta Metall. 36, 3131 共1988兲. 31 L. G. Rosenfeld, J. E. Ritter, T. J. Lardner, and M. R. Lin, J. Appl. Phys. 67, 3291 共1990兲. 32 A. Strojny, X. Xia, A. Tsou, and W. W. Gerberich, J. Adhes. Sci. Technol. 12, 1299 共1998兲. 33 D. Tabor, Hardness of Metals 共Clarendon, Oxford, UK, 1951兲. 34 M. Li, M. L. Palacio, C. B. Carter, and W. W. Gerberich, Thin Solid Films 416, 174 共2002兲. 35 X. Li and B. Bhushan, Acta Mater. 47, 473 共2002兲. 36 Polymer Handbook, edited by J. Brandrup, E. H. Immergut, and E. A. Grulke 共Wiley, New York, 1999兲, p. V/82. 37 Polymer Data Handbook, edited by J. E. Mark 共Oxford University Press, Oxford, 1999兲. p. 830. 38 Styrene, Its Polymers, Copolymer and Derivatives, edited by R. H. Boundy and R. F. Boyer 共Van Nostrand Reinhold, New York, 1952兲. 39 R. Feng and R. J. Farris, J. Mater. Sci. 37, 4793 共2002兲.

silicon interfaces

silicon interfaces

Suggest Documents