Final Report Annex 3 Enhancing propagation into buildings · ERA Report 2004-0072 Annex 3 7 1. Introduction This is Annex 3 to the Final Report provided under Ofcom Contract AY4464,

Page 1 of 54

ANTENNA SYSTEMS

Application of FSS Structures to Selectively Control the Propagation of signals into and out of buildings Annex 3: Enhancing propagation Into buildings

M Philippakis, C Martel, D Kemp

S Appleton

S Massey

ERA Report 2004-0072 A3 ERA Project 51-CC-12033 FINAL Report Client : Ofcom Client Reference : AY4464

ERA Report edited and checked by: Approved by:

Martin Shelley Project Manager

Robert Pearson Head of Antenna Systems

February 04Ref. Z:\AS_Projects\Custom Antennas and Consultancy_SW\12033_RA_in_and_out_building_FSS\Reporting\FINAL REPORTING\Annex 3 Passive FSS.doc

ERA Report 2004-0072 Annex 3

2

Crown copyright 2004. Applications for reproduction should be made to HMSO.

This report has been prepared by ERA Technology Limited and its team for the Ofcom under Contract No. AY4464.

DOCUMENT CONTROL

The document may be distributed freely in whole, without alteration, subject to Copyright.

ERA Technology Ltd Cleeve Road Leatherhead Surrey KT22 7SA UK Tel : +44 (0) 1372 367000 Fax: +44 (0) 1372 367099 E-mail: [email protected]

Read more about ERA Technology on our Internet page at: http://www.era.co.uk/


3

Contents

Page No.

1. Introduction 7

2. FSS requirements: spectral properties of illumination 7

3. Equivalent circuit analysis of FSS structures 10

3.1 The equivalent circuit method 11

3.2 Transmission through dielectric layers 14

4. Matching glazing with FSS 17

4.1 Cases analysed 17

4.2 Basic properties of glazing structures 17

4.3 Matching using grid FSS 18

4.4 Matching using “mostly metal” FSS 20

4.5 Manufacturing technologies 30

4.6 Regulatory issues 31

5. Matching cavity walls with FSS 32

5.1 Basic RF properties of brick walls 33

5.2 The effect of loss in the cavity wall model 37

5.3 Tolerance analysis of cavity wall model 39

5.4 Low cost manufacturing technologies 42

5.5 Regulatory issues 52

6. References 54


4

Figures list

Page No.

Figure 1: ‘Test room’ illuminated by an in-building dipole radiator 7 Figure 2: Electric field distribution due to a dipole radiator inside the ‘test room’ at 400 MHz 8 Figure 3: Electric field distribution due to a dipole radiator inside the ‘test room’ at 900 MHz 8 Figure 4: Angular distribution of incidence energy at f=400MHz 9 Figure 5: Angular distribution of incidence energy at f=900MHz 10 Figure 6: Equivalent circuit parameters for TE incidence on the plane T 11 Figure 7: Equivalent circuit parameters for TM incidence on the plane T 12 Figure 8: Equivalent circuits for periodic square loops 12 Figure 9: Equivalent circuit for grids 13 Figure 10: Equivalent circuit for double square loops 13 Figure 11: Equivalent circuit for gridded square loop 13 Figure 12: Equivalent circuit for gridded double square loops 14 Figure 13: Equivalent circuit for gridded Jerusalem cross 14 Figure 14: Transmission line model for double glazing 14 Figure 15: Reflection coefficient for TE and TM polarisation for a 60° incidence angle 16 Figure 16: Transmission coefficient for TE and TM polarisation for a 60° incidence angle 16 Figure 17: Transmission and reflection for float glass at 60° incidence 18 Figure 18: Transmission and reflection for double glazing (float) at 60° incidence 18 Figure 19: The effect on the S11 of float glass with and without a matching section 19 Figure 20: Predicted performance for float glass with a matching section 20 Figure 21: FSS and its complementary, (a) Inductive grid, (b) capacitive patches 21 Figure 22: Transmission line model for the FSS in Figure 21 (a) 21 Figure 23: Transmission characteristics of a grid and patch FSS 22 Figure 24: Single annular slot geometry 23 Figure 25: Transmission characteristics of a single annular slot FSS: TE incidence 24 Figure 26: Transmission characteristics of a single annular slot FSS: TM incidence 24 Figure 27: Dual annular slot geometry 25 Figure 28: Transmission characteristics of a dual annular slot FSS: TE incidence 26 Figure 29: Transmission characteristics of a dual annular slot FSS: TM incidence 26 Figure 10: Singular annular slot FSS 28 Figure 11: Double glazing measurement setup 29 Figure 12: Comparative performance of the double glazing 29 Figure 30: Brick geometry with dimensions in mm 33 Figure 31: Brick model in a periodic skewed lattice 34 Figure 32: Measured real part of the complex relative permittivity of brick and mortar 34 Figure 33: Measured electric loss tangent of brick and mortar 35 Figure 34: Transmission coefficient of the brick wall without mortar 36 Figure 35: Transmission coefficient of a mortar wall without brick 36 Figure 36: Transmission coefficient of the brick wall with mortar 36


5

Figure 37: Cavity wall model 37 Figure 38: Loss effects in the cavity wall model for TE transmission at 0° incidence 38 Figure 39: Loss effects in the cavity wall model for TE transmission at 60° incidence 38 Figure 40: TE transmission for cavity wall with Thermolite block rε = 3.2 40

Figure 41: TE transmission for cavity wall with Thermolite block rε = 3.55 40

Figure 42: TE transmission for cavity wall with Thermolite block rε = 3.9 40 Figure 43: TE transmission for cavity wall with Thermolite block thickness = 90 mm 41 Figure 44: TE transmission for cavity wall with Thermolite block thickness = 100 mm 41 Figure 45: TE transmission for cavity wall with Thermolite block thickness = 110 mm 41 Figure 46: QMP manufacturing process 43 Figure 47: PCB manufacturing process 44 Figure 48: FSS locations in a cavity wall 53

Tables list

Page No.

Table 1: Glass structures analysed 17 Table 2: Comparison of single and dual slot FSS performance 27 Table 3: Impact of using “mostly-metal” FSS on building regulations 31 Table 4: Cavity wall analysed 32 Table 5: Material properties used for frequencies around 400 MHz 35 Table 6: QMP Resolution 46 Table 7: Cost of screen printing system 48 Table 8: Cost of inkjet printing system 48 Table 9: Cost of electrolytic system 49 Table 10: QMP chemical costs 49 Table 11: Electrolytic plating chemical costs 50 Table 12: Substrate costs 50 Table 13: Cost of screen printing ink 50 Table 14: Cost of inkjet-printing ink 51 Table 15: Sheet cost for screen printing 51 Table 16: Sheet cost for ink-jet printing 52 Table 17: Impact of using cavity wall FSS on building regulations 53


6

This page is left intentionally blank


7

1. Introduction

This is Annex 3 to the Final Report provided under Ofcom Contract AY4464, Application of FSS Structures to Selectively Control the Propagation of signals into and out of buildings. It gives a detailed description of the work carried out concerned with propagation through glass and cavity wall structures.

In Section 2, an analysis is described which was aimed at identifying whether preferential angles of incidence are present in typical propagation scenarios. Section 3 describes the basic analysis technique that was used to undertake the bulk of the theoretical work described. Section 4 provides details of the work undertaken to improve propagation through glass. This focuses on transmission through K-glass, a coated material providing thermal insulation which is almost completely opaque at RF frequencies. Section 5 describes the work undertaken to assess the potential to improve radiation through cavity walls. References are provided in Section 6.

2. FSS requirements: spectral properties of illumination

Exterior walls and windows clearly form a barrier to efficient propagation of radio waves into and out of building and, hence, propagation which may be improved using FSS structures to match these elements. In general, the performance of FSS structures depends on the angle of incidence of the illuminating wave. In this section, a short study is described which aimed to identify if there are any more common angles of incidence on the walls or windows of the building so that structures can be optimised to reflect this.

A typical office type room with dimensions of 4.5 x 6 x 2.5m (width w, length l, height h) was considered (Figure 1). Six 1.1 x 1.25m openings, corresponding to windows, were included in the model.

4.5 m

6.0 m

2.5 m

Aperture plane

Dipole radiator

x

y

z

Figure 1: ‘Test room’ illuminated by an in-building dipole radiator


8

The theoretical analysis was based on the FEKO ray tracing modelling tool. Metal walls were assumed since the current version of FEKO cannot be used to consider dielectric sheets. A dipole type illumination was used. The radiator was placed at a number of different locations inside the ‘test room’ and the angles of incidence on the wall surfaces analysed.

Typical examples of radiated field distribution inside and outside the ‘test room’ can be seen in Figure 2 and Figure 3 for frequencies of 400 and 900 MHz.

Figure 2: Electric field distribution due to a dipole radiator inside the ‘test room’ at 400 MHz

Figure 3: Electric field distribution due to a dipole radiator inside the ‘test room’ at 900 MHz


9

In order to evaluate the spectral characteristics of the radiation that illuminates the walls, an “aperture plane” such as the one shown in Figure 1 was created. This is located slightly inside the room (aperture plane to wall separation 0.01m) to avoid coinciding exactly with the metallic walls. Using ray tracing, it is possible to evaluate the electric fields that are tangential to the aperture plane. Fourier Transformation (Plane Wave Spectrum) of the aperture fields will reveal if there are any preferred directions of illumination.

The analysis was conducted for a number of illumination scenarios. Typical 3D spectral distributions for the aperture field distribution at the interface plane can be seen in Figure 4 and Figure 5.

It is clear that there is no preferred directional trend. However, it has been shown that, for cases where the radiator is aligned to radiate vertically polarised waves, the bulk of radiation (≥ 70% of power) available in the aperture tends to be distributed within a cone with an apex angle of 60°. This was used as the limiting value when designing FSS structures.

Figure 4: Angular distribution of incidence energy at f=400MHz (a) y = +0.25 w from room centre along y-axis (b) y = at room centre (c) y= -0.25 w from room centre along y-axis

(a)

(c)

a)


10

Figure 5: Angular distribution of incidence energy at f=900MHz y = + 0.25 w from room centre along y-axis

y = at room centre y= -0.25 w from room centre along y-axis

3. Equivalent circuit analysis of FSS structures

Equivalent circuit modelling of FSS requires very limited computer resources when compared to full three dimensional electromagnetic modelling and is therefore useful for quickly predicting the performance of structures. The circuits also provide a useful physical insight into how an FSS works as its parameters are changed.

(a)

(c)

(b)


11

3.1 The equivalent circuit method

The starting point for developing equivalent circuits for FSS structures is the circuit representation of an infinite parallel conducting strip developed by Marcuvitz [Ref 1]. The development of the strip array formulation for TE incidence is shown in Figure 6. The metal strips have a zero thickness, a width, w , and period p . The plane wave is incident onto the strips at an angle q .

E

θ

w p

X

Zo Zo

Figure 6: Equivalent circuit parameters for TE incidence on the plane T

The equivalent circuit inductive reactance is calculated by:

( ) ( ) ( )

+

== θλ

πλθλ ,,,

2sin

1lncos,, wpG

pw

pwpFZwX

o

( )( ) ( )

( ) −+−+

−+−+

++

−++

−

++

−−

=AAAA

AAAAwpG

682

22

22

22

282

14

1

44

11

21,,,

βββββ

βββθλ

1sin

122

−

−

=±

λλθ pp

A

p2sin πωβ =

where l is the wavelength and oZ is the characteristic impedance of free space.

Similarly, the equivalent circuit representation for TM incidence is shown in Figure 7. The incident magnetic field vector is parallel to the metal strips and is incident at an angle of θ . The strips have a period, p , and a gap spacing g . The capacitive susceptance is calculated by:


12

( ) ( )4 , ,o

B gF p g

Zλ=

H

θ

gp

B

Zo Zo

Figure 7: Equivalent circuit parameters for TM incidence on the plane T

The equations presented above are valid for wavelengths and angles of incidence θ in the range

( )1 sin 1p θ λ+ < . They are also only valid for plane waves incident in either the E or H plane and

hence they cannot be used to model the cross polarisation effects of the FSS.

Equivalent circuits for different periodic structures can be obtained by modification of the strip array formula presented above. For example, a square loop FSS and its equivalent circuit are shown in Figure 8.

s

d

p

g

Zo ZoL

C

Figure 8: Equivalent circuits for periodic square loops [Ref 2]

Marcuvitz’s strip array formula can be adapted to account for the finite lengths of strips. The reactance of this structure is:

( ),2 ,L

o

X dL F p sZ p

ω λ= =

and the susceptance is given by:

( )4 , ,c

o

B dC F p gZ p

ω λ= =


13

The reactance and susceptance are each reduced by a factor d p from the corresponding array of infinite strips to account for the finite strip length. Furthermore, if the period is greater than the gap spacing, g p<< , the widths are equal to 2s .

Using the approach described above, equivalent circuits can be identified for other FSS elements. Examples of these structures are grids, the double square loop, the gridded square loop, the gridded double square loop and the gridded Jerusalem cross FSS, which are shown in Figure 9 to Figure 13.

Zo Zo

L1

Figure 9: Equivalent circuit for grids

Zo ZoL1

C1

L2

C2

Figure 10: Equivalent circuit for double square loops [Ref 3]

Zo ZoL1

C1L2

Figure 11: Equivalent circuit for gridded square loop [Ref 4]


14

Zo ZoL1

C1

L2

C2

L3

Figure 12: Equivalent circuit for gridded double square loops

Zo ZoL1

C1

L2 C2

Figure 13: Equivalent circuit for gridded Jerusalem cross

The formulations of these equivalent circuits do not take into account the effect of dielectric substrates. The presence of a dielectric substrate will only effect the capacitance of the circuit and not the inductance. The effects of dielectric layers can be accounted for by considering them as transmission lines connected to the equivalent circuits of the FSS. This method is discussed in the next section.

3.2 Transmission through dielectric layers

The analysis of plane waves travelling through dielectric layers has been treated comprehensively in literature [Ref 5] and therefore only a brief review of the concepts will be presented here. Consider the example shown in Figure 14, an equivalent circuit representation for double-glazing.

TLIN

F0=EL=Z0=ID=

fs GHzD[1]=17.05 DegZte[2]=143.4 OhmTL1

TLIN

F0=EL=Z0=ID=

fs GHzD[2]=12.97 DegZte[3]=377 OhmTL2

TLIN

F0=EL=Z0=ID=

fs GHzD[3]=17.05 DegZte[4]=143.4 OhmTL3

PORT

Z=P=

Zte[1]=377 Ohm1

PORT

Z=P=

Zte[5]=377 Ohm2

Glass GlassVacuum

Free space

Free space

Figure 14: Transmission line model for double glazing


15

The double-glazing analysed consists of two 6mm thick planes of glass separated by a 12 mm thick vacuum. These layers are represented as transmission lines in the model, which are characterised by their impedance and electrical length. The equations presented above can be used to calculate these parameters. The ganged transmission lines are connected to ports at each end to terminate the model with free space. In Figure 14, the impedance values shown are for an angle of incidence of 0°. In this case, the port terminations and the vacuum have the free space propagation impedance of 120π, or 377 Ω. The glass layers have an impedance of 143.4Ω and an electrical length of 17.05° at 800 MHz. As the angle of incidence varies, these parameters change accordingly.

An outline of the underlying physics is given below. A plane wave is incident on a dielectric layer at an angle of iq from normal. The dielectric constant of the sheet is re and the thickness is d and the

index of refraction is rh em e= » . The characteristic impedance of free space measured in a

direction at an angle iq to the direction of propagation of the plane wave is:

1 and cos coscos cos

o o oi o i

o i i o

ZZ Z Zm m q qe q q e^ = = = =P

where Z^ and ZP are the characteristic impedance for transverse electric (TE) and transverse

magnetic (TM) incident fields. The impedance of the wave within the dielectric are given by:

and coscoso o

rr r r

Z ZZ Z qe q e^ = =P

where rq is the angle of the propagating wave passing through the dielectric. This angle is calculated using Snell’s law of refraction and is given by:

1 sin sinsin where sin

i ir

r

q qq hh q

-= =

The electrical length of the dielectric with thickness d is:

( )2' cos rd dp ql

=

The equations presented above can be used to determine the impedance and electrical length of an arbitrary number of dielectric layers for TE and TM incident fields. Only the permittivity, thickness and angle of incidence onto the first dielectric layer are required to determine these relations which can be used to construct transmission line representations of the dielectric layers. Therefore, with knowledge of the incident angle and material properties, an equivalent circuit can be constructed to represent the propagation through the multilayer materials.

Commercially available software is used to construct these equivalent circuits. The software, Microwave Office, is a general-purpose linear and non-linear circuit analysis tool. It contains many “off the shelf” components for ease of building circuit models. This modelling technique has been


16

validated by comparing reflection and transmission data obtained using this approach with an ERA written code called GUIDE. GUIDE predicts the reflection and transmission through cascaded infinite dielectrics for arbitrary angles of incidence and has been extensively validated through measurements.

A comparison of the reflection and transmission through double glazing, based on the use of the equivalent circuit model and GUIDE, over a frequency range of 0-4 GHz and at an incidence angle of 60° is shown in Figure 15 and Figure 16. The results from the two methods are identical.

Figure 15: Reflection coefficient for TE and TM polarisation for a 60° incidence angle

Figure 16: Transmission coefficient for TE and TM polarisation for a 60° incidence angle


17

4. Matching glazing with FSS

4.1 Cases analysed

The materials listed in Table 1 below have been used to assess the potential for FSS structures to improve propagation through glass. In all cases, the initial equivalent circuit analysis was based on loss-less dielectric materials. Since float and toughened glass have very similar properties, all analysis has been carried out based on the properties of float glass. It has been assumed that low-emissivity glass (K-glass) is float glass with a solid metal coating, giving no transmission under any circumstances.

Table 1: Glass structures analysed

Building type Material profile Relative permittivity

Thickness (mm)

Float glass Glass 7.01 6

Toughened glass Glass 6.912 6

Double glazing (float) Glass

Vacuum

Glass

7.01

1

7.01

6

12

6

Double glazing (toughened glass)

Glass

Vacuum

Glass

6.912

1

6.912

6

12

6

4.2 Basic properties of glazing structures

Various glass structures were analysed for a number of incidence angles. The transmission and reflection coefficients for 60° incidence for 6mm float glass and double glazing (using 6mm float glass and a 12mm gap), are shown in Figure 17 and Figure 18 respectively. Being electrically thin, both the single layer and double layer glass structures have similar transmission performance at frequencies up to 2GHz. The double glazing has slightly higher losses which is expected since the plane wave travels through two layers of glass.


18

Figure 17: Transmission and reflection for float glass at 60° incidence

Figure 18: Transmission and reflection for double glazing (float) at 60° incidence

4.3 Matching using grid FSS

FSS can be used to match both the cases analysed above. The matching procedure is to add lumped elements to the transmission line models. The parameters for these lumped elements can be derived from the analytical models for the FSS. An initial study using different periodic structures was carried out and the grid configuration was chosen to match the structures. This structure can easily be implemented on both the glass and the wall without significantly affecting the cosmetic appearance.


19

The benefits of using a matching section based on a grid FSS can be seen in Figure 19. In this example, a simple grid of wires is placed on the surface of a single layer of float glass. The grid adds inductance to the transmission line model. This inductance can be chosen to provide optimum transmission for a given frequency. When the desired inductance is obtained, the physical dimensions of the FSS can be found from the equations presented above. In this case, the level of the reflection is reduced from 30% of the energy to less than 1% of the energy at 950MHz.

This process can also be performed in reverse. Therefore, the design of the matching section can be performed taking into account practical constraints on feature sizes. For example, to implement the matching section in Figure 19 a grid with a thickness of 35µm and a period of 248 mm is required; this grid thickness would be difficult to achieve so an alternative design may be preferred.

Figure 19: The effect on the S11 of float glass with and without a matching section

To prove the concept of matching with FSS, a design for the float glass was developed suitable for measurement in a waveguide (WG4). The waveguide has dimensions of 248 by 124 mm, and its operational frequency is from 0.75 to 1.12 GHz. To achieve a useful measurement, the periodicity of the grid must be a sub-multiple of the waveguide “a” dimension and manufacturing constraints defined the minimum thickness of the grid to be 150µm. These constraints are restrictive, making the design of an optimum matching section impossible; however, a reasonable theoretical performance was still obtained. Figure 20 shows the predicted performance of the matching section for 0° incidence. These predictions show the obvious benefits of using a matching section for glass. This design was manufactured by QinetiQ.


20

Figure 20: Predicted performance for float glass with a matching section

4.4 Matching using “mostly metal” FSS

Metal-coated glass has been identified as a structure for which an FSS would be highly beneficial. K-glass, which uses a very thin, optically transparent layer of sputtered Tin Oxide particles, is being more and more widely applied to buildings to improve the thermal efficiency of the building. Unfortunately, this layer forms an impenetrable barrier to RF propagation. At frequencies above about 2.5GHz, this can be an advantage, making the screening of Wireless LANs more straightforward. However, below 2.5GHz, transparency is required to allow the use of TETRA, GSM and UMTS applications inside buildings. The patch FSS and other “mostly-metal” structures, have significant potential to provide the required RF transparent characteristics at these lower frequencies without significantly degrading the thermal properties of the glass.

4.4.1 Equivalent circuit analysis of “mostly-metal” FSS

Marcuvitz [Ref 1] developed closed form expressions for the lumped impedances of loss-less infinite strips with zero thickness. The equivalent circuit representation of many FSS structures can be obtained by suitable modification of these infinite strip array formulas. However, these equations cannot be used to obtain the equivalent circuit representation of more complex FSS configurations using, for example, rings or spirals. Derivation of an equivalent circuit representation of these elements is possible, but the complexity of the analysis increases rapidly. Furthermore, certain assumptions must be made in the formulation, which reduce the accuracy of the solution. Therefore, as the intricacy of the element increases, equivalent circuit analysis must be replaced by the use of a full wave solver.

For FSS structures based on strip arrays without dielectrics, the complementary solution can easily be extracted. For example, the infinitely thin inductive grid FSS shown in Figure 21 (a) can be represented by the equivalent transmission line problem shown in Figure 22 [Ref 6].


21

(a) (b)

Metal

Non metal

Figure 21: FSS and its complementary, (a) Inductive grid, (b) capacitive patches

2Y

I T

R

Figure 22: Transmission line model for the FSS in Figure 21 (a)

The FSS is represented by a normalised shunt admittance 2Y. The reflection and transmission coefficients for this circuit can be calculated from simple transmission line theory as:

YT

+=

11

1)1(1

1−=

+−

= TY

R

which applies to both the inductive grid and a complementary structure consisting of square metallic patches, shown in Figure 21 (b). The relationships between transmission and reflection coefficients for the grid and patch elements can be obtained from Babinet’s principle [Ref 7] and are given by:

indcap RT −= and indcap TR −=

When the FSS is infinitely thin and there are no dielectrics, the performance of the patch exactly complements that of the grid. The transmission performance for a grid and a patch FSS under these conditions is shown in Figure 23. When the grid FSS is in its passband, the patch design is in its stopband, and vice versa.


22

Finite metal thicknesses and dielectric loading have an effect on the response of an FSS and its complementary solution and the expressions presented in the equations above are no longer valid and equivalent circuit analysis is no longer possible.

Figure 23: Transmission characteristics of a grid and patch FSS

4.4.2 FSS design for K glass

The work method of moments (MoM) technique was used to analyse “mostly-metal” FSS structures. Two “mostly-metal” designs, based on annular slots, were investigated for application to a double-glazed window.

4.4.2.1 Single annular slot FSS

The first design is based on single annular slot FSS which is shown in Figure 24 [Ref 8]. The double-glazed window is constructed from two panes of 6 mm float glass, separated by a 12 mm air gap. The glass was modelled using a permittivity of 7.01 and a loss tangent of 0.017. The metallic shielding of the K glass is implemented in the simulations as an infinitely thin sheet of metal on the inner face of one of the panes of glass. The FSS was designed using an optimisation procedure that uses the ring diameter and lattice spacing as variables. The design goal was to obtain an optimum transmission response at 900 MHz for incidence angles in the range 0° to 60°. The optimum design is shown in Figure 24. The FSS is based on tightly spaced slots approximately 25mm in diameter with a width of 0.25 mm. Using this geometry, more than 95% of the metal surface is retained.


23

60°

Glass

Glass

FSS

Slots

0.25

Vacuum

6

12

6

All dimensions in mm

Figure 24: Single annular slot geometry

The transmission characteristics for TE and TM polarised fields for 0° to 60° incidence angles are shown in Figure 25 and Figure 26 respectively. A benefit of this design is that the transmission passband is relatively broad. For both TE and TM polarisations, the transmission rapidly increases above 500 MHz. At 900 MHz, the worst-case transmission loss for the FSS is for TM polarisation at 60° incidence, which is 2.15 dB. Furthermore, at this frequency, the transmission response is reasonably constant with angle of incidence. The stability of the transmission loss with incidence angle tends to reduce away from the optimisation frequency. Even so, the worst-case loss is approximately 6 dB for TE polarisation at 1700 MHz. The equivalent losses for clear glass double glazing are 4dB for both 900MHz and 1700MHz.


24

Figure 25: Transmission characteristics of a single annular slot FSS: TE incidence

Figure 26: Transmission characteristics of a single annular slot FSS: TM incidence


25

4.4.2.2 Dual annular slot FSS

The single annular slot design is shown to have a large transmission bandwidth operating from 800 MHz to above 2GHz for small angles of incidence. For higher angles of incidence, the losses for TE polarised fields increase. These maximum losses occur close to the GSM band of 1800 MHz, so it would be desirable to reduce the losses at this band while retaining the performance at 900 MHz. A dual band structure is necessary to achieve this.

A dual annular slot FSS provides two resonances, one at a lower frequency (caused by the larger slot) and the other at a higher frequency (caused by the smaller slot) [Ref 9]. The dual slot FSS was designed using the same principles as for the single slot. The slot diameter and lattice spacing were used as variables in the optimisation procedure. The FSS design was optimised for best transmission performance at 900 and 1800 MHz over multiple angles of incidence. The final design is shown in Figure 27. The FSS consists of two concentric slots with a width of 0.25 mm, on a 60° lattice of about 35mm.

60°

Glass

Glass

FSS

Slots

0.25

Vacuum

6

12

6

All dimensions in mm

0.25

Figure 27: Dual annular slot geometry

The dual band transmission characteristics for TE and TM polarisations are shown in Figure 28 and Figure 29 respectively. An inherent characteristic of a dual resonance FSS is that a rejection band is formed between the passbands. This occurs in this design just below 1500 MHz. In the lower


26

passband region, the effect of angle of incidence shifts the passband peak to a lower frequency for TE, while the opposite effect occurs for the TM incident fields. The worst-case loss at 900 MHz is approximately 6 dB. In the upper passband, the sensitivity to incident angle is significantly reduced for values of less than 60°. The TM response remains nearly constant for all angles of incidence. The greatest loss at 1800 MHz is 4 dB for the TE polarisation at 60°.

Figure 28: Transmission characteristics of a dual annular slot FSS: TE incidence

Figure 29: Transmission characteristics of a dual annular slot FSS: TM incidence


27

A comparison of the transmission response at GSM and WLAN frequencies for an uncoated double glazed panel, and for double glazed panels incorporating both the single and dual slot FSS, is shown in Table 2; clearly, for K-glass double glazing, the transmission loss is always very high (typically more than -30dB). The greatest transmission losses are for TE polarised fields at high angles of incidence. The single slot FSS performs marginally better than the dual slot FSS at 900 MHz. Furthermore at 1800 and 2400 MHz, the dual slot FSS is better. These results indicate that a single or dual band complementary FSS can be designed to allow propagation of RF energy through normally opaque K-glass with varying degrees of efficiency.

Table 2: Comparison of single and dual slot FSS performance

Freq (MHz) Polarisation Incidence angle

900 1800 2400

Average loss (dB)

TE / TM 0 -1.24 -1.00 -0.29

TE 60 -3.82 -4.44 -1.44

Standard float glass double glazing

TM 60 -0.19 -0.37 -0.34

1.31

0 -1.65 -2.03 -3.92 TE 60 -1.02 -6.18 -3.8

0 -0.44 -2.43 -4.28

Single slot FSS + double glazing

TM 60 -2.15 -1.06 -2.37

2.61

0 -1.35 -1.23 -2.21 TE 60 -5.82 -3.97 -2.71

0 -3.12 -1.16 -2.39

Dual slot FSS + double glazing

TM 60 -0.22 -0.51 -1.99

2.22

4.4.3 Breadboard characterisation

A breadboard unit, using a 900 mm x 900 mm double glazed panel incorporating the single annular slot design, was manufactured. The FSS film was bonded onto one of the inside surfaces; the sample was not optically transparent. The panel was compared with two other double glazed units, one containing two panes of standard float glass and one containing a single pane of float glass and a pane of K-glass. All the units had the same overall dimensions and used the same glass and separation thicknesses.

The FSS panel was manufactured by QinetiQ in two sections. These were subsequently aligned and overlaid by ERA with both copper sides in contact to form an electrical contact; Figure 30 shows this join region of the two sections of the FSS. The FSS panel was bonded onto the inner pane of the glass using a spay mount adhesive and taped in the central join region to ensure the continuity of electrical


28

contact. Unfortunately, during the manufacture of the double glazed unit, which resulted in the FSS sheet being sealed inside, the tape was removed. During removal of the tape, some of the copper was stripped from the polyester substrate and the contact along the joint was degraded.

Figure 30: Singular annular slot FSS

The transmission properties of the panels were measured using a two-antenna transmission measurement. Two log-periodic antennas, connected to a vector network analyser, were positioned either side of the double-glazed panel, as shown in Figure 31. The transmitted signal was time gated to minimise the effects of edge diffraction and multipath. The glass measurements were referenced to the case when no glass was present.

The comparative measurement for the three panels and the theoretical prediction for the FSS panel are shown in Figure 32. The standard float glass had a maximum loss of 1.7 dB across the band, whereas the K glass was consistently more than 14 dB. The FSS structure significantly improved the transmission through the double-glazing when compared to the K-glass case; at resonance, the losses are broadly equivalent to those measured using the standard float glass unit. The measured resonance of the FSS panel is higher than the theoretical predictions. This shift between the measurement and theory is believed to be due to the delamination of the two panels. Note that the plot for clear glass goes above 0dB at low frequency. This is believed to be attributable to multipath effects; signals constructively add for the float glass at these frequencies.

From the results, it is clear that the FSS structure provides significant improvements to the RF performance of metal coated glazing. Optically opaque coatings as used in this experiment clearly cannot be used in a real building environment and it will be necessary, in a further programme of work, to look at the performance of optically transparent coatings to ascertain whether the same benefits can be achieved.


29

Figure 31: Double glazing measurement setup

Figure 32: Comparative performance of the double glazing


30

4.5 Manufacturing technologies

4.5.1 Modifying coated glass

The European Technical Centre of Pilkington has provided the following input, concerning the technical and commercial feasibility of implementing “mostly-metal” FSS structures on glass.

4.5.1.1 Base materials

Pilkington manufactures coated glass using two different generic processes.

“On-line” coating is a term which describes the process of introducing a coating while the glass is being manufactured. K-glass is one of a number of products which is manufactured using the “on-line” process. This uses a Tin Oxide layer.

“Off-line” coating refers to the process of putting down a coating as a separate process after the glass has been manufactured. A wide variety of coatings can be applied using this process. In particular, an layer of Silver can be deposited, sandwiched between two “matching” dielectric layers to make the glass optically transparent, which provides significantly better conductivity than the Tin Oxide used in K-glass.

4.5.1.2 Technical Feasibility of patterning the coatings

It would be possible to etch the required ring FSS pattern on coated glass which is manufactured using either “on-line” or “off-line” processes. Etching using a CO2 laser would be the preferred approach. This process is already used locally on some automotive products to create an RF window in a metal-coated windscreen to allow the use of screen-mounted electronic tags used by drivers using toll motorways.

Pilkington already uses lasers on its production lines to scan glass for imperfections at line speeds of 1m/sec (for a 6m x 3.2m panel). In principle, they believe it would be feasible to introduce a laser etching system into the line which could also create the required pattern at the same production line speeds.

4.5.1.3 Costs

Pilkington was unable, at the meeting, to assess the capital and unit costs of introducing the process into a glazing line with any degree of accuracy. It was felt that the unit manufacturing cost may increase by 10-20%. No estimate of capital costs was provided. It was noted that the price of the patterned glass would depend as much on market forces as on production costs.

4.5.1.4 Markets

Low E glazing was being used in 90% of new office installations in Germany, but in only 10% of installations in the UK. To date, Pilkington had not been made aware of any problems with the use of wireless telecommunications into buildings which had been directly attributed to the fitting Low E


31

glass. However, it was recognised that such an association may not be obvious to the operators of the buildings. Pilkington expressed interest in becoming involved in further research to develop glass that combined the benefits of thermal insulation with the frequency selective RF transparency, but stated that it was unlikely to take such technology to the marketplace until it was mandated by legislation.

4.5.2 Alternative manufacturing approaches

An alternative approach would be to coat ordinary glass with a polymer film. Multi-layer non-metallic optical interference films are available which can be tuned to reflect radiation in a particular wavelength band. A film could therefore be developed which was relatively clear in the visible region but reflective in the infrared, say 950 – 1150 nm. There is, however, a limit to the amount of reflection that can be obtained in this way. Greater reflection can be achieved by adding a metallic coating to the film. Such a film could serve two purposes:

• forming the FSS structure and • improving the U-value of the unit compared with clear glass.

A further alternative for glazed facades would be to make use of the spandrel panels. These are often made from glass without a metallic coating, backed by insulation and plasterboard. They could be made to look similar to the transparent glazing of the windows by painting the inner surface of the glass. A polythene layer could be used as a vapour barrier. Provided there is no metallic layer, this should be reasonably transparent to radiation. Pilkington produce such spandrel panels to match the appearance of many of their glass types.

4.6 Regulatory issues

Etching circular ring patterns into K-glass and similar low-emissivity glasses will reduce the coated area by up to 10 %. This may have implications on the thermal management of buildings, particularly as the regulations covering glazing structures become ever more stringent. Table 3 provides a summary of the expected impact on each of the Approved Documents.

Table 3: Impact of using “mostly-metal” FSS on building regulations

Reg Pass/Fail Comment

A - K N/A

L Special measured needed to ensure compliance

Thermal transmission of window would be increased, see notes (1) and (2) below

M - N N/A

Workmanship & buildability

Pass Coated glazing need to be handled with care to avoid damaging the coating.


32

1. Removal of metallic coating from the glass will increase the thermal transmission. The U-value of the window will increase in proportion to the amount of sputtered material removed. A window with a wood or PVC frame has a U-value of 2.0 W/m2K, as presently required by Part L, with the following glazing parameters: a) interpane gap at least 16 mm, b) air filled, c) emissivity 0.15 (K-glass).

If it is argon filled, about 10% of the glass area could have the coating removed, to give the same U-value of 2 W/m2K. Alternatively if a soft coating is used, emissivity 0.05, again about 10% of the glass area could be un-coated, to give the same U-value. Note that the currently proposed “Mostly-metal” FSS only has 5% metal removal. Removal of larger areas of coating would require further changes to the window such as addition of a third pane; triple glazing with clear glass meets the current Building Regulations Part L requirements.

2. Future revisions of Part L are likely to see a tightening of thermal insulation requirements. Thus a system which just meets current requirements for thermal transmission could fall outside any new tighter requirements. However, as outlined in above, the addition of films and use of a triple glazed unit may offset the losses in performance.

5. Matching cavity walls with FSS

The transmission characteristics of a cavity wall are dependent on the dielectric parameters of the individual wall materials, their thickness and their relative locations. In this section, the effect of loss and variation of the electrical permittivity are investigated. The cavity wall assessed here is constructed using a layer of brick, followed by an insulating material (modelled as air) and a layer of Thermolite block. Measurements on samples of brick and Thermolite block have shown that their electrical permittivities vary significantly from sample to sample and also that they are dependant on moisture content. Therefore, it is considered that effective matching of cavity walls using FSS will be very challenging.

Table 4 below shows the wall build used to assess the potential for FSS structures to improve propagation through a cavity wall.

Table 4: Cavity wall analysed

Material profile Relative permittivity Thickness (mm)

Brick 3.5 – 5.5 100

Insulator 1 100

Thermolite block 3.2 – 3.9 100


33

5.1 Basic RF properties of brick walls

This section describes the theoretical study undertaken to assess the RF propagation through typical brick wall section, taking in account the finite size and non-rectangular shape of the bricks and the mortar in between.

Figure 33 shows the geometry of a typical brick. The brick dimensions were measured from a real brick sample. Figure 34 shows the unit cell used to model an infinite brick wall. The lattice is skewed so that the relative displacement between each layer of bricks can be accounted for. When used, the mortar completely fills the void between the bricks. The mortar gap is assumed to be 10 mm (dimension measured from the ERA main building).

170

63

24

100

100

210

72

Top view

Side view

Figure 33: Brick geometry with dimensions in mm


34

Figure 34: Brick model in a periodic skewed lattice

The material proprieties of the bricks and the mortar, which were measured by NPL, are used to carry out the theoretical assessment. Figure 35 and Figure 36 show the measured relative permittivity and the loss tangent of brick and mortar respectively. Unfortunately, material property data is not available in the frequency region of 400 MHz. The measured permittivity data has therefore been extrapolated from these measurements to give the values shown in Table 5.

1

1.5

2

2.5

3

3.5

4

4.5

5

0.40 0.50 0.60 0.70 0.80 0.90 1.00 1.10 1.20

Frequency (GHz)

Rea

l (Er

)

Brick

Mortar

Figure 35: Measured real part of the complex relative permittivity of brick and mortar


35

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0.40 0.50 0.60 0.70 0.80 0.90 1.00 1.10 1.20

Frequency (GHz)

Loss

tang

ent

Brick

Mortar

Figure 36: Measured electric loss tangent of brick and mortar

Table 5: Material properties used for frequencies around 400 MHz

Relative permittivity Loss tangent

Bricks 4.45 0.031

Mortar 2 0.725

A selection of wall types was analysed at various frequencies including 400 MHz and 900 MHz for plane wave incidence angles ranging from 0° to 70°. The transmission loss as a function of frequency and incidence angle was calculated for the following cases:

• a homogeneous brick wall (Figure 37), • a homogeneous mortar wall (Figure 38), • a realistic brick wall with mortar (Figure 39).

In general, the transmission degrades as the angle of incidence increases. The transmission degradation becomes significant for incident angles higher than 60°. Solid mortar has losses between 0.5 and 2.5 dB more than homogenous brick. The transmission characteristics of a realistic brick wall lie between those for a solid brick and a solid mortar construction, but do not seem to follow the same well behaved trends (ie linearly degrading over frequency) as the solid dielectrics. Therefore, it appears not to be possible to accurately model a brick and mortar wall using an “effective dielectric”.


36

-20

-18

-16

-14

-12

-10

-8

-6

-4

-2

0

0 10 20 30 40 50 60 70Incident angle (deg)

Tran

smis

sion

(dB

)

350 MHz400 MHz450 MHz850 MHz900 MHz950 MHz

Figure 37: Transmission coefficient of the brick wall without mortar (TE polarisation)

-20

-18

-16

-14

-12

-10

-8

-6

-4

-2

0


Tran

smis

sion

(dB

)


Figure 38: Transmission coefficient of a mortar wall without brick (TE polarisation)

-20

-18

-16

-14

-12

-10

-8

-6

-4

-2

0


Tran

smis

sion

(dB

)


Figure 39: Transmission coefficient of the brick wall with mortar (TE polarisation)


37

5.2 The effect of loss in the cavity wall model

In this section, the wall build shown in Figure 40 is analysed with loss included. The ohmic loss places a fundamental limit on the performance improvement that can be realised by the addition of a matching section. In this analysis, the grid matching section has been omitted so that the true effects of the material losses are clear; reflection losses are still included. The loss tangents used in the model are based on an average of the measurements undertaken by NPL. Mean permittivity values were used when assessing losses.

Figure 40: Cavity wall model

Transmission losses through the wall are highest when the incident plane wave is TE polarised; therefore, the analysis is shown for TE polarisation only. The transmission performance with and without losses is compared in Figure 41 and Figure 42, for 0° and 60° incidence respectively. For information, frequencies relating to TETRA, GSM, UMTS and WLAN are also shown in the figures. At both incidence angles, the inclusion of loss systematically reduces the transmission through the wall. As expected, the loss increase is greater at higher frequencies (greater electrical thickness) and for largest angles of incidence. As an example, at 60° incidence the incorporation of loss results in a reduction in transmission of close to 10 dB.

The ohmic loss places a fundamental limit on the performance improvement that can be realised by the addition of a matching section.

BRICK (εr=4.5, tanδ= 0.03)

θ

INSULATOR (εr=1, tanδ= 0.00)

THERMOLITE BLOCK(εr=3.6, tanδ= 0.085)


38

Figure 41: Loss effects in the cavity wall model for TE transmission at 0° incidence

Figure 42: Loss effects in the cavity wall model for TE transmission at 60° incidence


39

5.3 Tolerance analysis of cavity wall model

The electrical properties of the building materials used in a cavity wall can vary quite considerably as a function of moisture content, in particular, and also from sample to sample. The physical properties of such materials are also not as well maintained, as would be the case for materials traditionally used in microwave applications. In this section, the effect of varying the permittivity and thickness of the wall are analysed for a lossless system. The cavity wall model shown in Figure 40 is used as the baseline. The permittivity of the brick and the Thermolite block is varied over the range of the measurement values presented earlier, and the thickness is altered by 10 % of its nominal value of 100 mm.

A square grid matching section is included in the model. The FSS is designed to improve the transmission characteristics around the TETRA and GSM bands at 400 and 900 MHz respectively. The optimum design is a square grid with a lattice period of 216 mm and grid thickness of 1 mm. This FSS is placed on the outer surface of the brick.

Losses are greatest for TE polarised fields at large incidence angles. Therefore, 60° incidence has been used in the assessment, and the results for TM polarisations are excluded for clarity. The effects of varying the permittivity of the brick layer between 3.5 and 5.5, using fixed values for the dielectric constant of the Thermolite layer of 3.2, 3.55 and 3.9, are shown in Figure 43, Figure 44 and Figure 45.

It can be seen that changing the permittivity of the brick has a significant effect on the predicted transmission bands of the wall. An increase in the permittivity reduces the resonance frequencies of the transmission bands at the TETRA and GSM1 bands. The effect is even more significant at the GSM2 band. Similar levels of degradation are seen when the Thermolite layer properties are changed.

The effects of varying the thickness of the brick layer between 90 and 110mm, using fixed values for the thickness for the Thermolite layer of 90, 100 and 110mm, are shown in Figure 46, Figure 47 and Figure 48.


40

Figure 43: TE transmission for cavity wall with Thermolite block rε = 3.2




41

Figure 46: TE transmission for cavity wall with Thermolite block thickness = 90 mm




42

These results highlight the major impact that changes in electrical and physical properties of building materials have on the RF performance of a cavity wall. The passband shift is less pronounced in the TETRA band, but due to the high-Q response at this frequency, even a small shift can result in a large change in transmission loss. At 900 MHz, where the passband bandwidth is larger, reasonable shifts in resonance frequency due to material changes can be accommodated. At frequency bands higher than this, the effects are less predictable.

These results indicate that it is very difficult to design an FSS that can provide consistent performance for a number of different cavity wall geometries. A more appropriate technique would be to have a reconfigurable FSS that can be tuned to optimise the transmission characteristics for predefined passband frequencies. A number of these active FSS structures have been addressed in WP3400 and further work is on-going to assess the potential to use two layer passive FSS structures which are moved with respect to each other and the wall.

5.4 Low cost manufacturing technologies

5.4.1 Introduction

This section described a novel low cost manufacturing technology that has been developed by QinetiQ, which could be ideally suited to the implementation of FSS structures in the built environment.

QinetiQ Metal Printing (QMP) is an innovative process for creating patterned metal deposits, using a range of metals and alloys, onto most rigid and flexible substrates. In contrast to alternative methods, QMP is an efficient and cost-effective route to patterned metals and, because the original electrical and RF properties of the base metal are preserved, it lends itself well to the creation of frequency selective surfaces (FSSs).

The QMP process affords considerable reductions in operating capital, equipment costs and waste materials compared to traditional acid etch processes. In addition, the use of non-toxic chemicals in the QMP process provides an environmentally safe alternative to those employed in processes today.

5.4.2 The QMP Process

The QMP process in its simplest form is a tool which can be used to produce metal patterns on a range of substrates. QMP have invented a catalytic ink (Patent GB 01 134 08, GB 01 285 71) that can be deposited by a range of printing techniques to form patterned metals, for example to form FSS structures. The focus of this current project is on inkjet- and screen-printing.

Once the ink has been printed onto a substrate by one of these printing methods, it is dried at about 80-100 °C. The substrate, in this case either polymer or glass sheet, is then immersed in an electroless solution. Metal is grown on the printed ink regions, with negligible sideways growth (typically 1%). Although the thickness grown varies from metal to metal, a few µm of coating is typical. Once plating is complete the substrate with its metal coating is rinsed and dried.


43

5.4.3 Benefits of the QMP process

The benefits of the QMP process in a wide-range of applications are significant and many. This section identifies those benefits, and highlights the advantages of this technique over conventional production methods, such as lithography and etching.

5.4.3.1 Manufacturing

Typically, the QMP process will enable the following savings:

• Capital costs - 90% saving. • Floor space - 90% saving. • Waste - 99% saving. • Materials - 75% saving.

These figures are from a direct comparison between a reel-to-reel QMP system and a conventional flexible printed circuit board (PCB) production line with a web width of 610mm.

The QMP process uses only five steps to produce an FSS pattern on a polymer (Figure 49); this compares to the 30 or more required during conventional PCB manufacture (Figure 50). Restricting the number of processing steps dramatically reduces the cost of the capital equipment and the manufacturing floor area that it occupies.

Figure 49: QMP manufacturing process

Print QMP

Dry Copper growth

Rinse/Dry FSS WasteSubstrate

=


44

Figure 50: PCB manufacturing process

QMP is an additive processes (i.e. metal only grows where the ink is printed) and is extremely economic in terms of material waste. Indeed, the only waste generated is during the replacement of the electroless chemicals, which takes place every 4-6 months. This waste is of low toxicity and is readily disposed of.

The manufacture of FSS for buildings may require a wide web width. The widest webs used for reel to reel PCB processing are 610mm wide, this being driven by the size of the lithographic development units. QMP does not have the same constraints so the web width is only constricted by the width of available polymer. FSS produced by a QMP system could be 2-5m wide, depending on polymer availability.

The QMP process typically deposits 1-2µm of copper, but if thicker metal is required, an electrolytic stage can be easily added to the production line. A limitation is that a pattern can only be electroplated if it is continuous.

If a metal thickness of more than 10µm were required, then etching of copper-clad substrates would provide a more cost affective method. Note that web widths would be a limiting factor in this case.

Stage 1: Obtain copper-coated PCB substrate

Stage 2: Apply etch resist in required pattern

Rinse / Dry

DevelopExpose pattern

Soft bake Apply photo-resist

Hard bake

Stage 3: Selective etch

Substrate SnCl2 treat

Electroless Copper

Rinse / Dry

PdCl2 treat

RinseRinse Dry

Rinse / Dry

Etch copper

PCB Remove resist

Waste Rinse / Dry


45

5.4.3.2 Substrates

There is a diverse and wide range of surfaces to which the QMP process is applicable. The process can be applied to virtually any water-resistant material, such as synthetic paper, polyester, polypropylene, polyimide, FR4, ceramic, and glass.

The wet chemistry stage of QMP prevents it being used to directly create FSS on some building materials, such as breeze block and plasterboard. However, QMP can be used to create FSS patterns on polymer films that can, in turn, be attached to such materials. In the case of vapour shield plasterboard, which has an aluminium backing film, it may be possible to replace the aluminium film with QMP copper on a polymer substrate.

5.4.3.3 Application method

QMP is unique in that it can use of wide variety of print methods. It can be printed using industrial “drop on demand” inkjet or desktop inkjet printers. Such digital printing methods are ideal for fast prototyping and limited production where a quick turnround of candidate patterns is required. However, for mass production of a fixed FSS pattern, other printing methods such as gravure, flexographic or rotary screen-printing are more suitable. These methods have all been demonstrated in the context of QMP, and they all offer wide web widths, high resolutions and potentially very rapid print speeds in industrial-sized systems.

Wide-web inkjet systems are more specialised and are normally bespoke for a particular customer, at inherent additional cost. Inkjet-printed QMP patterns are less stable when electroplated. Work is being performed to resolve this problem, but currently their use is not recommended for patterns requiring electroplating.

5.4.3.4 Metal surfaces

A range of metals can be grown, including copper, nickel, cobalt, iron, tin, silver and gold. Mixed magnetics can also be deposited, for example 45% nickel, 5% boron, 25% iron, 25% cobalt. These blends are described as mixes rather than true alloys, as they are the results of co-deposition rather than a heating process.

In addition, a combined QMP copper/nickel mix is currently being used for EMP shielding by a leading defence contractor who has replaced metal boxes with QMP coated plastics.

5.4.3.5 Performance

There is a wide range of EMP shielding inks/pastes and conductive polymer products available. QMP differs from these conductive pastes and polymers in a number of ways.

Conductive inks rely on a high loading of a conductive metal, usually silver, in a non-conductive binder. Conduction is caused by contact between silver particles within the binder structure. However, such inks have a much higher RF resistance and do not match the performance of a true


46

metal. In some pastes, the silver particle loading is so high that the ink price mirrors the price of the raw silver component. Conductive polymers have lower dc and ac conductivity.

The metals deposited by QMP are as pure as chemical deposition allows. Copper is typically 98% pure, nickel is 90%. QMP metals are fully dense and exhibit the same RF characteristics as a pure metal skin.

5.4.3.6 Quality

QMP has an internal QA system for guaranteeing its inks from batch to batch. Metal coatings have been produced which satisfy a wide range of criteria. EMP coatings satisfy UK MOD and American Mil Standard Peel Strength tests.

5.4.3.7 Resolution

The resolution achievable using QMP is solely dependent on the print process used. A list of the print processes and currently achievable resolutions can be found in Figure 6 below.

It should be noted that QMP is also suitable for the printing of double-sided aligned FSS patterns, since the first ink coating can be dried before the second coating is applied. Both ink surfaces are then cured before electroless plating. The tolerance of the alignment process is also listed.

Table 6: QMP Resolution

Print Process Resolution (µm) Double sided tolerance (µm)

Inkjet Industrial 140 Theoretical 140

Inkjet Desktop 70 N/A

Screen-print (sheet) 200 200

Screen-print (rotary) 150 150

Gravure 70-90 70-90

During the electroless plating stage the metal grows on all of the surfaces of the ink, giving both sideways and vertical growth. If a 2µm thick copper pattern is deposited there will typically be an additional 2µm of metal on the sidewall of each track. This is not ideal but is significantly better than the results caused by undercut in the etching process.

5.4.3.8 Prototyping

QMP offers a rapid prototyping capability for small-scale samples, using desktop printing inks. The maximum sample size is currently limited to A4 sheets, but this can easily be increased depending on customer requirements. Digitally produced A3 to A0 sheets could feasibly be produced for the evaluation of FSS designs. It should be noted that the desktop printers can only print on a synthetic


47

paper and that generally these patterns cannot be electroplated, limiting this facility to producing 1-2µm copper and 1-10µm nickel patterns.

5.4.3.9 Scalability

The simplicity of QMP is that it is a new production method based on existing production machinery. Hence, scaling-up of QMP would use existing machinery from the wide-format printing industry in conjunction with QMP ink, followed by metal growth using industry standard electroless- and electro-plating systems.

In full-scale production the web width of a reel-to-reel QMP line would only be restricted by the available polymer web width. Referring back to section 5.4.3.2, a QMP system with a web width of 2-5 m using a rotary screen- or gravure-printing stage would be relatively easy to produce.

5.4.3.10 Customisation

QMP has a range of stock inks, which have been tailored for customer’s individual substrates. These can be modified to customers needs, this process normally taking 1-2 weeks, depending on the substrate.

5.4.4 Cost estimates

5.4.4.1 Introduction

The cost estimates in this section are broken down into the following categories:

• Printing systems for the deposition of the QMP ink, • Electrolytic chemistry system for the subsequent thickening of the QMP metal, • Consumables such as chemicals and polymer substrate, • QMP inks, • Indicative yearly costs for production by each method.

Although the QMP inks can be deposited by a variety of methods, this study has concentrated on screen- and inkjet-printing because these are known to achieve the typical pattern resolutions likely to be needed for polymer- and glass-backed FSS structures. Further costs have been provided for subsequent electrolytic plating, should this be necessary, to thicken the QMP metal coatings.

Costs have been prepared for printing systems and electrolytic chemistry systems capable of producing three web widths: 600, 1200 and 2400 mm.

After the initial investment in the production facility, the cost of the FSS sheets is dictated by the price of the QMP ink, the substrate polymer and the consumables of the production line. The cost of inks is driven by volume, hence this portion of the cost estimates has been based on the area coverage of the FSS design. Bulk prices have been estimated for the chemicals and polymer substrates.


48

5.4.4.2 Assumptions

The following assumptions have been made in preparing these cost estimates:

• A standard thickness of 1-2µm of copper is assumed for these calculations, which is typical of the QMP process,

• The reel-to-reel linear speed is 3m/min, • Prices are for new equipment.

5.4.4.3 Hardware Costs

5.4.4.3.1 Screen-printing system

The estimated initial Capital Equipment cost for a reel-to-reel screen-printing system with electroless (QMP) deposition facility is shown below as a function of web width. The printing system would use metal mesh screens (typically 200 mm in diameter) and would also include a corona treatment stage for preparation of the substrate, electronic across web registration to prevent pattern drift and either hot air or infra-red dryers to cure the QMP ink. An in-line electroless deposition stage and further dryers complete the system.

Table 7: Cost of screen printing system

Web Width (mm) Price (£K)

600 95

1200 135

2400 175

5.4.4.3.2 Inkjet-printing system

The initial Capital Equipment cost for a reel-to-reel inkjet-printing system with QMP deposition is shown below as a function of web width. The price includes corona treatment, drying and in-line electroless QMP deposition, and is based on the same basic structure as the screen-printing system.

Table 8: Cost of inkjet printing system


600 105

1200 155

2400 195


49

5.4.4.3.3 Electrolytic System

An in-line electroplating system would require specially designed anode and cathode arrangements and control circuitry, suitable for maintaining the correct electrical current through the FSS elements during plating. The resistance of any one FSS circuit would create the need for a given current. Additional chemicals would be needed to maintain the purity of the deposited metal.

The transit time of the web in the plating bath is dictated by the rate of coating, typically 20µm/hour. Running at 3m/min approximately 200m of substrate would be required and a coating tank capable of holding such a length of polymer.

Table 9: Cost of electrolytic system


600 65

1200 75

2400 100

5.4.4.3.4 Consumables

Chemicals

The yearly price for QMP chemicals, based on bulk buying and three bath regenerations, are shown in Table 10 below. The chemicals are more expensive than those used for simple electroplating (Table 11).

Table 10: QMP chemical costs

Web Width (mm) Tank capacity (litres) Chemical costs (£K)

600 6000 6

1200 12000 12

2400 24000 24

The yearly prices of electrolytic plating chemicals, based on bulk buying, are shown below for various web widths. No bath regenerations are necessary for electroplating.


50

Table 11: Electrolytic plating chemical costs

Web Width (mm) Tank capacity (litres) Chemical costs (£K)

600 600 0.4

1200 1200 0.8

2400 2400 1.2

Substrates

The costing has been performed for 50µm thick polyester film. Bulk buying of quantities >500 kg can reduce the price by up to another 25%.

Table 12: Substrate costs

Web Width (mm) Cost per Sheet (£)

600 x 600 0.36

1200 x 12000 1.15

2400 x 2400 3.40

Inks

The QMP ink for polyester is in the process of commercialisation and a final purchase price has not yet been set. The prices stated in the following tables are indicative only.

Table 13: Cost of screen printing ink

Coverage (%) FSS Sheet Size (mm) Ink Price (£)

10 0.16

50 0.78

90

600 x 600

1.40

10 0.64

50 3.12

90

1200 x 1200

5.6

10 2.56

50 12.48

90

2400 x 2400

22.4


51

Table 14: Cost of inkjet-printing ink

Coverage (%) FSS Sheet Size (mm) Ink Price (£)

10 0.02

50 0.19

90

600 x 600

0.33

10 0.08

50 0.76

90

1200 x 1200

1.32

10 0.32

50 3.04

90

2400 x 2400

5.28

5.4.4.3.5 Yearly costs for FSS Production

The following tables give estimates of total yearly output of QMP FSS sheets and the costs per sheet (consumables and inks only) as a function of FSS area coverage. Although the figures are for reel to reel web processing, an effective sheet size is used to show how the price per sheet decreases as the sheet size increases. It should be noted that the range of potential FSS coverage areas is very wide. For simple grid structures it is likely to be well below 5%. For “mostly-metal” structures, it could reach 95%.

Screen-printing

Table 15: Sheet cost for screen printing

Coverage (%)

Sheet Size (mm) Number of Sheets Cost per sheet (£) QMP only

Cost per sheet (£)Electroplating

10 0.55 0.01 50 1.17 0.01 90

600 x 600 720000

1.79 0.01 10 1.86 0.01 50 4.34 0.01 90

1200 x 1200 350000

7.22 0.01 10 6.10 0.01 50 16.02 0.01 90

2400 x 2400 175000

25.94 0.01


52

Inkjet-printing

Table 16: Sheet cost for ink-jet printing

Coverage (%)

Sheet Size (mm)

Number of Sheets

Cost per sheet (£) QMP only

10 0.41 50 0.58 90

600 x 600 720000

0.72 10 1.3 50 1.98 90

1200 x 1200 350000

2.54 10 3.86 50 6.58 90

2400 x 2400 175000

8.82

5.4.5 Comments

The QMP process is extremely scalable. Even with the small scale equipment costed above, a large quantity of product can be produced cheaply. It should be noted that the dominating recurring cost is the per sheet cost of the polymer substrate; this could be reduced by buying in bulk.

The price of the electrolytic chemicals is so low that in production quantities the additional cost per item is negligible and £0.01 is quoted as a maximum.

There is a significant difference in ink price between inkjet and screen print inks. This is caused by the amount of ink used in a screen print compared to an inkjet print.

5.5 Regulatory issues

For the purposes of assessing the regulatory impact of fitting FSS structures into walls, it has been assumed that the FSS consists of a rectangular grid of square metallic patches printed onto a plastic or polymer substrate. This can either be attached to the insulation batts for installation in the cavity or used as ‘wallpaper’ on the inside of the wall. It is assumed the external face of the wallpaper presents a suitable medium for decorating (painting or additional decorative wallpapering). Possible locations for the FSS film are shown in Figure 51 below.


53

Figure 51: FSS locations in a cavity wall

Table 17 provides a summary of the expected impact on each of the Approved Documents.

Table 17: Impact of using cavity wall FSS on building regulations

Reg Pass/Fail Comment

A N/A No structural implications

B Pass See note 1

C Pass Vapour permeability needs to be designed accordingly (see note 2)

D Pass Assuming the FSS is metal element is sealed and washable (see note 3)

E - K N/A

L Pass Likely to increase the thermal resistance (see note 4)

M - N N/A

Workmanship and buildability

Pass Recommend placing FSS on inside wall (see note 5)

1. This would seem to introduce no more fire risk than any of the other materials found in a cavity, inner leaf or inside wall. Thin facings (0.5 mm or less) do not generally affect the fire classifications, so walls and ceilings can effectively be papered. Use of thicker laminates needs to be substantiated by flame propagation tests.

2. The FSS would introduce a further impermeable membrane within the cavity, though continuity could not be guaranteed and therefore it would not be advisable to assume it has the same function as a vapour barrier. For the inside installation it would be advisable to give the FSS substrate a similar characteristic with regard to vapour permeability as wallpaper to enable the wall to breathe and new plaster to dry.

Inside Outside


54

3. An issue is any moisture run-off from the FSS will have to comply with any regulations about metals in waste water. Thus, the use of FSS on the outside of the building will raise issues about runoff. Increasing the amount of metal in the cavity may raise similar issues.

4. Whilst the installation is unlikely to alter the thermal characteristics of the wall the membrane should not be considered as a continuous thermal barrier especially if installed in the cavity.

5. From the perspective of construction it would be possible to put the FSS inside the cavity, say, by printing the FSS on to the rigid insulation batts. These will have to conform to the appropriate sizes for installation between the wall ties and the system will not be continuous. There is a distinct possibility that the batts may be damaged or incorrectly installed. Also they will become inaccessible once put in place. Use of the FSS on the inside of the building would enable ease of fitting, continuous placement and correctly placed FSS. This obviously could be retro-fitted to buildings. This would require similar properties to wallpaper and the metallic elements would need to be reasonably sealed.

6. References

[Ref 1] Marcuvitz, N., Waveguide Handbook Radiation Laboratories Series, McGraw Hill, 1951

[Ref 2] Langley, R. J. and Parker, E. A., Equivalent circuit model for arrays of square loops Electron. Lett., 18, pp. 294-296, 1982

[Ref 3] Hamdy, S. M. A. and Parker, E. A., Current distribution on the elements of a square loop frequency selective surface Electron. Lett., 18, pp. 624-626, 1982.

[Ref 4] Langley, R. J. and Parker, E. A., Double square frequency selective surfaces and their equivalent circuit Electron. Lett., 19, pp. 675-677, 1983

[Ref 5] Collin, R.E., Field Theory of Guided Waves McGraw Hill, 1960

[Ref 6] Lee, S. W., Zarrillo, G., Law, C. L., Simple Formulas for Transmission Through Periodic Metal Grids or Plates IEEE Trans. AP, vol 30, no. 5, Sept. 1982, pp 904 – 909.

[Ref 7] Lo, Y. T., Lee, S. W., Antenna Handbook. Theory Applications and Design Van Nostrand Reinhold, 1988, pp 2-13 – 2-16

[Ref 8] Parker, E. A. and Hamdy, S. M. A., Rings as elements for frequency selective surfaces' Electron. Lett., 17, pp. 612-614, August, 1981.

[Ref 9] Parker, E. A., Hamdy, S. M. A. and Langley, R. J., Arrays of concentric rings as frequency selective surfaces Electron. Lett., 17, pp. 880-881, November 1981