DesignCon 2016...DesignCon 2016 Effect of conductor profile structure on propagation in transmission...

21
DesignCon 2016 Effect of conductor profile structure on propagation in transmission lines Allen F. Horn III, Rogers Corporation Associate Research Fellow, Lurie R&D Center, PO Box 157 Rogers, CT 06263 USA [email protected] Patricia A LaFrance, Rogers Corporation Sr. Engineering Assistant [email protected] Christopher J. Caisse, Rogers Corporation Sr. Engineering Assistant [email protected] John P. Coonrod, Rogers Corporation Technical Marketing Manager [email protected] Bruce B. Fitts, Rogers Corporation Staff Development Engineer [email protected]

Transcript of DesignCon 2016...DesignCon 2016 Effect of conductor profile structure on propagation in transmission...

  • DesignCon 2016

    Effect of conductor profile structure

    on propagation in transmission lines

    Allen F. Horn III, Rogers Corporation

    Associate Research Fellow,

    Lurie R&D Center,

    PO Box 157

    Rogers, CT 06263 USA

    [email protected]

    Patricia A LaFrance, Rogers Corporation

    Sr. Engineering Assistant

    [email protected]

    Christopher J. Caisse, Rogers Corporation

    Sr. Engineering Assistant

    [email protected]

    John P. Coonrod, Rogers Corporation

    Technical Marketing Manager

    [email protected]

    Bruce B. Fitts, Rogers Corporation

    Staff Development Engineer

    [email protected]

    mailto:[email protected]:[email protected]

  • 2

    Abstract

    Designers of high frequency electrical devices have long known that conductor surface

    roughness affects loss. Earlier correlations such as Morgan under predict insertion loss

    by a large margin at higher frequencies and on narrower transmission lines where

    conductor effects dominate. The present work experimentally demonstrates that the

    recent Hall-Huray “snowball” model and the Sonnet conductor loss correlation correctly

    predict the shape of the insertion loss versus frequency curve up to 110 GHz on treated

    (nodulated) conductor surfaces. Quantitative agreement, however, requires empirical

    adjustment of the surface area index (SAI), snowball radius or RMS roughness. It is also

    shown that the untreated (not “nodulated”) surfaces of reverse-treated foils have little

    effect on increasing conductor losses, even though they exhibit a significant RMS

    roughness and SAI.

    Authors’ Biographies

    Allen F. Horn, III, Associate Research Fellow, received a BSChE from Syracuse

    University in 1979, and a Ph. D. in Chemical Engineering from M.I.T. in 1984. Prior to

    joining the Rogers Corporation Lurie R&D Center in 1987, he worked for Dow Corning

    and ARCO Chemical. He is an inventor/co-inventor on 16 issued US patents in the area

    of ceramic or mineral powder-filled polymer composites for electronic applications.

    Patricia A. LaFrance, Engineering Assistant, has 25+ years’ experience in the formulation

    and testing of composite materials for industrial and electronic applications. She joined

    the Rogers R&D Center in 1997.

    Christopher J. Caisse is a Senior Engineering Assistant at the Lurie R&D Center. Prior to

    joining Rogers he worked at Lucent Technologies and OFS Optics. He has spent the past

    12 years developing high frequency circuit materials, automated test software, and

    electrical and thermal test methods. Chris is working towards a BSEE at UCONN.

    John Coonrod is a Technical Marketing Manager for Rogers Corporation, Advanced

    Connectivity Solutions. John has 26 years of experience in the PCB industry. The first

    half of his career was spent on flexible circuit design, processing, and materials

    engineering. The past 13 years he has worked on high frequency circuit fabrication,

    application support, and conducting electrical characterization studies of Rogers HF

    laminates. John has a BSEE degree from Arizona State University.

    Bruce Fitts received a BS in Chemical Engineering from Northeastern University in

    1972. He served as a member of the corporate R&D Group of the Rogers Corporation

    from 1972 to 1993 formulating polymeric composite materials. From 1993 to 2002, he

    was the product development manager for the Moldable Composites Division of Rogers

    Corporation, and continued to work with Perstorp and Sumitomo after the sale of the

    division. In 2006, Bruce re-joined the R&D staff at the Advanced Connectivity Solutions

    division where he has worked on the development of polymer composite laminates for

    advanced printed circuit applications.

  • 3

    Introduction and Summary

    As operating frequencies have increased and dielectric thickness has decreased for both

    high speed digital and RF planar circuits, the effects of conductor profile on both

    attenuation and propagation constant have become more evident. In the present work we

    have characterized the roughness parameters Rq (RMS), Rz (peak-to-valley), and surface

    area index (SAI) of both the treated and untreated side of copper foils by white light

    interferometry. The SAI is the ratio of surface area of the structured surface to the

    smooth base area. The nodule radius of the copper treatment was estimated by SEM

    analysis. For the treated sides of the various foil samples, the measured Rq values ranged

    from 0.2µ to 3.0µ, SAI values from 1.0 to 3.9, and nodule radii from zero to 0.9µ.

    For the untreated (nodule-free) sides, the measured Rq values ranged from 0.2µ to 1.3µ

    and SAI values from 1.0 to 2.2.

    Microstrip fifty-ohm transmission lines were fabricated on 0.002” to 0.007” (0.05 mm to

    0.2 mm) ultralow loss dielectric laminates. Standard laminates were made with the

    treated (nodulated) side facing the dielectric. These materials exhibited an increase in

    insertion loss with increasing profile, as measured by SAI and Rq.

    A number of the reverse-treated foils exhibited sufficient adhesion on the untreated

    (nodule-free) side to allow fabrication of test circuits with the untreated side facing the

    dielectric. For this case, increasing Rq and SAI had little effect on the insertion loss. The

    lack of nodules and relatively long range periodicity of thickness variation leads to much

    lower energy dissipation. This is inherent in the Hall-Huray model.

    Comparison modeled with measured data show very good agreement of the shape of the

    insertion loss curve with Hall-Huray conductor model and the Sonnet correlation of

    surface inductance to foil profile. Due to the ambiguity in measuring nodule diameter,

    small adjustments to the parameters are required for quantitative agreement with

    relatively low profile foils.

    Quantitatively matching the insertion loss curves of the two high profile dendritic foils,

    with a structure quite different than the nodulated foils requires larger adjustments for

    quantitative agreement.

    Background

    It has been known since the early days of radar development using waveguides that rough

    surfaces will increase the conductor loss. In 1949, S. P Morgan1 published the results of

    numerically modelling the effect of rectangular and square grooves in a conductor with

    an aspect ratio of about 1 to 1. For a signal traveling perpendicular to the groove

    direction, the conductor loss would increase by a maximum of a factor of two with

    increasing frequency as the skin depth approached and then became smaller than the

    groove height. The increase in modelled conductor loss when the signal ran parallel to the

  • 4

    grooves was much smaller. Similar results were obtained in 1996 by Groisse et al2. by a

    finite element method

    The Morgan correlation was adapted into an automated microstrip insertion loss and

    impedance calculation described by Hammerstad and Jensen3 (H&J). The correlation is

    incorporated as a multiplicative correction factor KSR to the attenuation constant

    calculated for a smooth conductor.

    α cond, rough = α cond, smooth · KSR (1)

    where α cond, smooth is the attenuation constant calculated for a smooth conductor and

    2

    4.1arctan2

    1 KRMS

    SR

    R (2)

    where RRMS is the RMS value of the conductor roughness and δ is the skin depth. It

    should be noted that both α cond, smooth and KSR are functions of frequency. When the

    ratio of RRMS/ δ is small, as with a smooth conductor or at low frequencies where the skin

    depth is large, the value of KSR is close to one. As the ratio becomes large with higher

    profile conductors and higher frequencies, the value of KSR approaches two. This

    correlation predicts a “saturation effect,” i.e., that the maximum effect of the conductor

    roughness would be to double the conductor loss. This result also implies that the

    conductor loss for a lower profile foil will eventually approach that of a rough foil as

    frequency increases.

    Over the years, the Morgan correlation has been used with a reasonable degree of success

    and is standard text book knowledge4. However, as higher frequencies and thinner

    circuits with narrow conductors have become more prevalent, inaccuracies in the model

    have become evident.

    Conductor loss models

    In the last decade, a number of authors have examined the effect of conductor roughness

    on the propagation of signals in PCD-based transmission lines.5-9

    Brist et al5 and Liang

    et al6 used the Morgan correlation (equation 2) to achieve a causal model of laminate

    performance that agreed well with measured data up to 20 GHz. Hinaga et al7 used a

    similar correlation to obtain more accurate dielectric loss values. Chen8 used numerical

    EM modeling of a rough conductor with electroless nickel-immersion gold plating and

    achieved good agreement with measured data. Tsang et al9 have performed numerical

    and analytical simulations that show that for multi-scale rough surfaces (in contrast to the

    periodic surfaces treated by Morgan), saturation does not occur and increases of greater

    than a factor of two in conductor loss can occur.

  • 5

    “Multi-GHz, causal transmission line modeling methodology with hemispherical surface

    roughness approach” was proposed by Hall et al. “Fundamentals of a 3-D “snowball”

    model for surface roughness power losses” by Huray et al describes the approach of

    calculating the increase in attenuation factor due to the copper foil treatment modeled as

    spheres and the model is described in greater detail in later work.

    The simplified Hall-Huray model is applied as a roughness correction factor to the

    calculated conductor attenuation factor for smooth foil.

    α cond, HH rough = α cond, smooth · KHH (3)

    where KHH is given by

    (4)

    and SAI is the surface area index and a is the nodule radius as described in detail by

    Griesi et al. Griesi et al also discuss a number of methods of improving the

    characterization of the foil surfaces with optical image analysis and mechanical

    profilometer of the untreated surfaces of the base foil. The modeled data agreed very

    well with measured on a microstrip TL with a relatively smooth copper foil exhibiting an

    RMS surface roughness of 1.2µ, an effective nodule radius of 0.63 µ, and an SAI of 1.77.

    Simonovich also discusses an alternative method of determining modeling parameters for

    the Hall-Huray method. The predicted values of attenuation factor for the Hall-Huray

    and H&J methods for three of the copper foils studied is shown in figure 1.

  • 6

    A number of authors14-20

    have observed an increase in the propagation constant, phase

    length, and group delay that appears as an increase in extracted permittivity when the

    substrate properties are calculated from circuit performance. Deutsch et al14

    noted a 13%

    increase in extracted permittivity for an unspecified high profile foil compared to a low

    profile foil on 2.6 mil thick laminate from 250 MHz to 2 GHz. More recent work16

    recorded an approximate 15% increase in the extracted permittivity of a 50 ohm TL on 5

    mil very low loss laminate with 3µ RMS ED foil compared to that clad with 0.4 µ RMS

    rolled foil. Other works have noted increases of 2 to 5% extracted permittivity on thicker

    laminates with higher profile foils. The earlier models of conductor profile effects1-13

    do

    not address the observed increase in propagation constant and extracted permittivity.

    A model based on correlating an increase in surface impedance with measured RMS

    roughness has been incorporated into Sonnet Software and has been found to yield good

    agreement between measured and modeled data for both attenuation and propagation

    constants17-18

    . Schlepnev et al19

    proposes that the increase in extracted permittivity is

    due to increased capacitance caused by the copper foil dendrites. Koledintseva et al20

    propose an “effective roughness dielectric” where the local permittivity and DF are

    effected by the copper foil dendrite structure. The values of the permittivity, DF, and

    thickness of the “ERD” were extracted from measured foil roughness parameters and the

    measured attenuation and phase constants of circuits made with the various foils. The

    values of the ERD layer are the reinserted in a 3D EM model to calculate circuit

    performance arriving at good agreement between modeled and measured values.

    Current Work

    In the current work, we have characterized the surface of a wide range of copper foils

    using white light interferometry (WLI) and scanning electron microscopy (SEM). The

    smoothest of surfaces were nearly perfectly smooth with surface area indices of less than

    1.1 and no visible nodules. The untreated surfaces of reverse-treated (RT) foils, though

    having no nodules, exhibited RMS roughness values of greater than 1µ and SAI values of

    greater than two. The highest profile foils exhibited SAI values approaching 4 and a

    range of nodule sizes from 0.3 to 1 microns.

    The foils were laminated to several thin (0.05 to 0.2 mm) low loss dielectric materials.

    Microstrip fifty ohm transmission lines were etched and the differential insertion loss

    and phase length were measured to frequencies as high as 110 GHz. We were able to

    fabricate and test a number of circuits with the untreated surface facing the dielectric, to

    measure the effect of the different surfaces independently.

    It is found that the RA foil behaves like perfectly smooth foil, in spite of having an SAI

    value of greater than 2 and 0.3µ nodules that are clearly visible by SEM. The untreated

    side of the RT foils also exhibit loss values approaching those of perfectly smooth foil,

    again in spite of measured RMS roughness values greater than 1µ and SAI values of

    greater than 2. It is implicit in the Hall-Huray model that these nodule-free surfaces

    exhibit very low loss. We show experimentally that the Hall Huray model fits the shape

    of insertion loss curves out to 100 GHz. However, with higher profile surfaces, the

  • 7

    modeling parameters need to be adjusted significantly from what is measured by SEM

    and WLI.

    Experimental methods and results

    Microstrip laminate samples

    Fifty-ohm microstrip transmission lines were photo-lithographically etched onto thin low

    loss high frequency circuit laminates. The majority of the samples ranged from 0.002” to

    0.007” in thickness. Thin samples of low loss dielectric materials maximize the

    contribution of the conductor to signal attenuation and propagation.

    Rogers ULTRALAM® 3850HT LCP (liquid crystal polymer) makes an excellent test

    vehicle for circuit properties. This material is a glass fabric-free, pure resin circuit

    substrate that relies on the inherently low CTE of the oriented LCP film to achieve a good

    in-plane CTE match to the copper foil. Since the ULTRALAM 3850HT substrate

    consists of a single pure substance, the variation in the dielectric properties is inherently

    low and there is no question of “glass to resin ratio” affecting the dielectric properties.

    When tested by a variety of methods, the dielectric constant of the LCP is 3.05 and the

    dissipation factor is 0.002 at 10 GHz.

    Rogers RO3003™ laminate is a silica-filled PTFE composite exhibiting a dielectric

    constant of 3.0 and DF of about 0.001 at 10 GHz. Rogers RO1200™ laminates are glass

    fabric reinforced silica-PTFE composites targeted for ultralow loss digital applications,

    with a dielectric constant of 3.0 and DF of about 0.0015 at 10 GHz.

    The majority of our samples were electrically characterized up to 50 GHz using long and

    short straight 50 –ohm transmission lines held in an Intercontinental W-7000

    Intercontinental Universal Substrate Fixture. The system was SOLT calibrated to the

    cable ends. Full S-parameter measurements were made with a 50 GHz Agilent PNA on

    long and short TLs. Insertion loss was calculated from the difference in S21 of the long

    and short TLs divided by the length difference. Differential phase length was calculated

    similarly to yield the effective permittivity, Keff of the microstrip circuit and the substrate

    permittivity, K’sub was extracted variously using Rogers’ impedance calculator, “MWI”

    based on the method of H&J, HFSS, or Sonnet Software.

    To obtain the 110 GHz data, we used 1mm Southwest Microwave model # 2492-04A-6

    end launch connectors with a grounded co-planar launch to 50 Ω microstrip TL. The

    lines were tapered in the launch area to improve the impedance match. Calibration was

    done using a 1mm mechanical kit with offset shorts SOLT for a single sweep calibration

    from 33 MHz to 110 GHz. S-parameter data were collected with a Keysight N5251A

    mmWave PNA, with 3303 testing points, IF Bandwidth set uo 300 Hz, and the power

    level at -5 dBm.

  • 8

    Profile measurements

    Planer organic composite circuit substrates are clad with one of three types of

    commercially available copper foil specifically manufactured for that purpose: rolled

    annealed, (RA), electrodeposited (ED) and reverse treated (RT). The foils are treated by

    the foil manufacturers with different types of treatments to improve and preserve

    adhesion to different types of circuit substrates. Historically, high profile (“rough”) foils

    have been used to increase adhesion to the dielectric material while lower profile foils

    are used to improve etch definition or reduce conductor loss.

    The surface profiles in the current work have been characterized using a Veeco

    Metrology Wyko® NT1100 optical profiling system. The instrument’s operation is

    based on white light interferometry (WLI). This non-contact method generates a three

    dimensional image of the surface topography with a resolution of about 1 nm in a 1 mm

    square area. The profile can be characterized by a wide variety of different statistics,

    including Rz, the peak-to-valley roughness, Rq (or RRMS), the root-mean-square

    roughness, and the surface area index (SAI). RRMS is most widely used in characterizing

    conductor roughness in high frequency electrical applications. However, Rz values based

    on mechanical profilometers are still widely reported by foil manufacturers. A

    representative WLI trace is shown in figure 2.

    The SAI is the measure of visible surface area to the underlying flat smooth surface. It

    should be noted that only the visible surface area is counted by the interferometry

    method. A checkerboard square with side 2a will have an SAI of (π/2 + 0.215) when

    covered by a hemisphere of radius a (the 0.215 due to the uncovered corners). The same

    square would exhibit a true SAI of (π + 1) when covered by a perfect sphere, but only the

    SAI of (π/2 + 0.215) would be visible when viewed head-on from above. Since the SAI

    is an essential component of the Hall-Huray simplified calculation, we were hopeful

    nonetheless, that this parameter could be helpful in calculating conductor loss.

    Figure 2

  • 9

    Copper foil cladding

    RA (rolled annealed) foil is produced from an ingot of solid copper by successively

    passing it though a rolling mill. After rolling, the foil itself is very smooth, with an RMS

    profile Rq of 0.2 to 0.3μ. For printed circuit substrate applications, the foil manufacturer

    additively treats the rolled foil, increasing the RRMS to 0.4 to 0.5 μ on the treated side, as

    measured by WLI. The SAI value is 2.1 by WLI and the nodule radius was estimated as

    0.3 based on SEM. The WLI SAI value for the untreated side of the RA foil is 1.1.

    Currently, commercially available RA foil is limited to a width of 26” (0.66 m).

    ED (electrodeposited) foil is produced by plating from a copper sulfate solution onto a

    slowly rotating polished stainless steel drum. The “drum side” of ED foil exhibits an

    RRMS of about 0.1 to 0.2μ, similar to untreated RA foil. The SAI of the untreated drum

    side of ED foils are less than 1.2. The profile of the “bath side” of the plated foil is

    controlled by the plating conditions, but is considerably higher in profile than the drum

    side. The ED foil manufacturer generally applies a further plated treatment to the bath

    side of the foil for improved adhesion and chemical compatibility with the intended

    dielectric material. As observed by Griesi et al, the treatment nodules deposited on the

    bath side of the foils tend to concentrate on the peaks of the underlying rough surface,

    giving rise to a dendritic appearance. The cross section in figure 3 demonstrates just how

    pronounced the dendrites can be. ED foils have historically been manufactured with

    RRMS values in the range of 1 to 3μ. The SAI values of standard ED foils with high

    profile range from 3 to 4 or even greater. These high values seem believable from simple

    inspection of the scanning electron micrographs (fig. 6) and the cross section of figure 3.

    Depending on the treatment type, we have observed nodules with radii ranging from 0.4

    to 0.9 μ.

    RT (reverse-treated) foil was developed to allow the manufacture of much lower profile

    from an electro-deposited based foil. The plating conditions are adjusted to make the

    bath side of the foil as smooth as possible. The adhesion promoting treatment is applied

    to the drum side of the base foil can result in a nodular structure similar to the treatment

    on RA foil. In our experience, the Rq values for treated side of RT foil are typically 0.7

    to 1.2μ and the SAI values range from about 2.2 to 3. We have measured nodule radii

    from 0.4 to 0.5μ on the RT samples.

    The untreated bath side of the RT foils can exhibit RRMS and Rz values that are higher

    than those of the treated side of the same foil. On the three RT foil types that we have

    measured SAI of the untreated bath side is also rather high for an untreated surface, in the

    range of 2.1 to 2.2. However, as is evident from comparing the SEM photos, the

    structure of the surfaces of the RT-MP treated and untreated sides are very different

    (figures 4 & 5). Even though the Rq of the untreated side is higher than the treated, its

    effect on conductor loss is much lower due to the lack of nodulation.

    Due to the growth of the lithium ion battery market, “battery foil” (BF) is now produced

    in large quantities by ED foil suppliers. The battery foil is made to be a as smooth as

    possible on both the bath and drum side. The two BF samples evaluated in the present

  • 10

    work were quite different from one another, with one exhibiting an extremely low profile

    treatment (BF-ELP) and the other, a relatively high profile treatment (BF-HP2).

    However, both exhibit low Rq and SAI on the untreated sides.

    We also evaluated an experimental ED foil, EXP, that exhibits similar properties of a

    measured SAI of 1.0 and Rq of 0.2µ on both sides. The structure is quite different than

    any other foil (fig. 6).

    The properties of the treated sides on 10 different foil samples are summarized in table 1

    and those of the untreated sides in table 2. RT denotes a reverse-treated foil and ED is

    for standard ED, treated on the bath side. LP and MP denote low and medium profile.

    All samples are ½ oz./ft2 (17 μ thickness) with the exception of RT-MP-1.0 which is 1

    oz./ft2. HP1 denotes one type of high profile treatment, and the HP2 treatment is a

    different type of treatment. Note that the samples include three different base foils, an

    RT, an ED, and a BF all with the same HP2 treatment.

    Scanning electron micrographs at a magnification of 7000X of most of the above foils are

    shown in figures 4-6. These photos were used to estimate the nodule radius. Not only is

    there a fairly wide distribution of radii within each material as discussed in detail by

    Greisi et al12

    , but there is clearly some ambiguity as to how to define a radius, particularly

    with the dendritic foils of figure 6.

    Figure 3 - Cross section of ED-HP1 foil

    Material Rq

    (um) Rz

    (um) SA

    Index

    Nodule Radius a (um)

    EXP 0.2 1.3 1.0 --- BF-ELP 0.3 2.1 1.1 0.2 Rolled 0.4 3.7 2.1 0.3 RT-LP 0.8 5.2 2.2 0.5

    RT-MP-0.5 0.8 4.2 2.3 0.5 RT-MP-1.0 0.9 4.1 2.3 0.5

    BF-HP2 0.7 4.8 2.7 0.4 RT-HP2 1.2 8.7 3.0 0.4 ED-HP1 3.0 15.0 3.4 0.9 ED-HP2 2.2 13.1 3.9 0.4

    Table 1 - Treated side

    Material Rq (um) Rz (um) SA Index

    EXP 0.2 1.3 1.0

    BF-ELP 0.3 2.1 1.1

    Rolled 0.3 2.1 1.1

    RT-LP 0.5 3.7 1.7

    RT-MP-0.5 0.8 5.8 2.2

    RT-MP-1.0 1.3 8.8 2.2

    BF-HP2 0.3 2.7 1.2

    RT-HP2 0.9 6 2.1

    ED-HP1 0.4 2.8 1.2

    ED-HP2 0.4 2.8 1.2 Table 2 - Untreated side

  • 11

    Figure 4

    Figure 5

  • 12

    Figure 6

    Results and Discussion

    An interesting characteristic of the insertion loss versus frequency curve for RO3003

    laminate (figure 7) is that the loss of the Rolled foil matches the modeled value for

    smooth foil, in spite of the fact that the Rolled foil exhibits significant nodular roughness,

    with an Rq of 0.4µ, SAI of 2.1, and nodule radius of about 0.3 µ. The authors have

    consistently observed this phenomenon over the last 20+ years.

    The measured foil parameters for ED-HP2 (blue line in figure 7) are an Rq value of 2.2µ,

    SAI of 3.9, and nodule radius of 0.4 µ. The Morgan/H&J correction calculated using an

    Rq value of 2.2 µ clearly grossly under predicts the insertion loss beyond 25 GHz.

    Sonnet, using the “thick metal” model, matches the data very well across the full range of

    110 GHz, with an input Rq value of 2.6 µ, rather than the WLI-measured value of 2.2 µ.

    Similarly, if the Hall-Huray model is an excellent fit across the full range, using the WLI-

    measured SAI value of 3.9, but reducing the input value of nodule radius, a to 0.2 µ.

  • 13

  • 14

    Similar results are obtained with the Rogers RO1200 laminate. The Rolled foil clad

    sample nearly perfectly matches the values modeled for smooth foil up to 100 GHz. The

    measured insertion loss of the RT-HP2 foil, with a measured Rq of 1.2µ, SAI of 3.0 and

    nodule radius of 0.4µ matches the HH model well, using the SAI of 3.0, but adjusting the

    input nodule radius to 0.15µ. Similarly, the Sonnet thick metal model fits the full range

    of measured data well if the Rq value is adjusted to 2.0 µ.

    The insertion loss versus frequency curve for 50 ohm transmission lines on 2, 4 and 7 mil

    thick ULTRALAM 3850HT laminates (figure 9) further demonstrates that the insertion

    loss with rolled foil matches that modeled for perfectly smooth with good accuracy.

    In figure 10 we compare the measured and modeled insertion loss curves for the highest

    profile (ED-HP1 and ED-HP2) foils. In this case, good agreement of the Hall-Huray

    model was obtained using the SEM-measured values of the nodule radius for each foil

    and adjusting the input value of SAI down to 2.0. Without this adjustment, the HH

    model would predict substantially higher loss than was measured. Given that the

    dendritic structure of these foils does deviate significantly from implied nodular structure

    of the HH snowball model, this is not surprising.

  • 15

    The importance of the detail structure and not just a single parameter such as Rq is shown

    by the difference in insertion loss of the treated (nodulated) and untreated RT-MP 1 oz/ft2

    and ½ oz/ft2 foils (figure 11). It was our good fortune that the untreated sides of most of

    the RT foils samples exhibited sufficient adhesion to make laminates and etch circuits to

    compare the performance independently of the two sides. Comparing the WLI-measured

    parameters in tables 1 and 2, it is seen that the Rq values of the untreated side is the same

    as the treated side for the ½ oz/ft2 foil, with for the 1 oz. foil the untreated side exhibits a

    higher Rq. However, the insertion loss values with the untreated sides facing the

    dielectric are considerably lower than with the nodulated side facing the dielectric.

    The comparatively small effect of Rq and SAI of the untreated, nodule-free side of the

    foil facing the dielectric is demonstrated for five samples of RT foils by a plot of the

    measured IL at 50 GHz versus Rq (figure 12). The measured IL increases from 1 dB/inch

    to 1.75 dB/inch with the nodulated side facing the dielectric with a slope of 0.8 and R2 of

    about 0.9 With the untreated side facing the dielectric, the IL at 50 GHz is flat, averaging

    about 1.1 dB/inch, and slope and R2 value of 0.1

    These results are implicit in the Hall-Huray model’s strong dependence of loss upon

    nodule radius. In the absence of nodules, there is very little effect of the conductor

    surface on conductor loss. An excellent simplified explanation of how the conductor

    surface texture affects loss has recently been given by Loyer21`

    .

  • 16

  • 17

    The nodulated surfaces also show a greater effect on the phase constant, and the value of

    the dielectric constant of the substrate that is extracted from the phase length. The details

    of the differential phase length calculation are explained elsewhere17,18

    .

    A plot of the extracted K’ of the substrate from phase length data using the method of

    H&J is shown for 50 ohm transmission lines on 0.004” LCP dielectric with RT-LP foil,

    both with treated and untreated side facing the dielectric (figure 13). In this case the

    nodulated side increases the value of extracted dielectric constant by about 0.1.

    The effect of foil nodulation on the propagation constant on the ten foils included in this

    study is shown in a plot of the value of the extracted K’ of the substrate versus Rq for

    both the treated and untreated conductor sides facing the dielectric. When the

    treated/modulated side is facing the dielectric, the extracted K’substrate value increases by

    as much as 15% with increasing Rq (red squares). When the untreated side is facing the

    dielectric, there is very little change in propagation constant with Rq (black diamonds).

  • 18

    Smooth foil options

    As discussed earlier, we have observed that rolled foil exhibits conductor loss

    approaching that of what is modeled for smooth foil at frequencies of up to 110 GHz. In

    addition to the limitation of only exhibiting good adhesion to certain dielectric materials,

    the use of rolled foil in laminate manufacture is limited by its maximum available width

    of 26” (66 cm). While many copper clad laminates for RF applications are made in

    relatively small formats, larger size laminates are required for lower cost applications.

    Copper foil manufacturers are well aware of the market for lower loss foils. The profiles

    of two of the ED foils included in this study, BF-ELP and EXP were lower than that of

    the rolled foil. As shown in figure 15, the insertion loss up to 50 GHz is very close to

    what is modeled for smooth foils. We are hopeful that smooth foil performance, with

    reliably good adhesions and favorable economics will be achievable on larger laminates

  • 19

    Conclusion

    In the present work, we have experimentally investigated the effects of

    surface structure on ten copper foil samples with a wide range of properties.

    In spite of exhibiting a SAI of 2.1 and a nodulated surface, rolled foils

    exhibits conductor losses matching what is modeled for smooth foil at

    frequencies of up to 110 GHz. ED foils with the lowest profiles exhibit

    similar insertion loss characteristics.

    Though the untreated (nodule-free) sides of reverse-treated copper foils can

    exhibit Rq and SAI values higher than that of the treated sides, they exhibit

    lower loss. As is implicit in the Hall-Huray model, the nodules are the

    major contributor to excess conductor loss.

    Foil nodules also affect the propagation constant, resulting in an increase in

    the phase length and extracted dielectric constant. The nodule-free surfaces

    exhibit much less of an effect.

  • 20

    References

    1. S. P. Morgan, “Effect of surface roughness on eddy current losses at microwave frequencies,” J. Applied Physics, p. 352, v. 20, 1949

    2. E. Hammerstad and O. Jensen, “Accurate models of computer aided microstrip design,” IEEE MTT-S Symposium Digest, p. 407, May 1980

    3. S. Groisse, I Bardi, O Biro, K Preis, & K. R. Richter, “Parameters of lossy cavity resonators calculated by the finite element method,” IEEE Transactions on

    Magnetics, v. 32, n. 3, May 1996

    4. D. M. Pozar, Microwave Engineering, 2nd Edtion, Wiley (1998) 5. G. Brist, S. Hall, S. Clouser, & T Liang, “Non-classical conductor losses due to

    copper foil roughness and treatment,” 2005 IPC Electronic Circuits World

    Convention, February 2005

    6. T. Liang, S. Hall, H. Heck, & G. Brist, “A practical method for modeling PCB transmission lines with conductor roughness and wideband dielectric properties,”

    IEE MTT-S Symposium Digest, p. 1780, November 2006

    7. S. Hinaga, M., Koledintseva, P. K. Reddy Anmula, & J. L Drewniak, “Effect of conductor surface roughness upon measured loss and extracted values of PCB

    laminate material dissipation factor,” IPC APEX Expo 2009 Conference, Las

    Vegas, March 2009

    8. X. Chen, “EM modeling of microstrip conductor losses including surface roughness effect,” IEEE Microwave and Wireless Components Letters, v. 17, n.2,

    p. 94, February 2007

    9. L. Tsang, X. Gu, & H. Braunisch, “Effects of random rough surfaces on absorption by conductors at microwave frequencies, IEEE Microwave and

    Wireless Components Letters, v. 16, n. 4, p. 221, April 2006

    10. S.H. Hall, S.G. Pytel, P.G. Huray, D. Hua, A. Moonshiram, G. Brist, and E. Sijercic, "Multi-GHz, Causal Transmission Line Modeling Methodology with a

    Hemispherical Surface Rough-ness Approach", IEEE Transactions on Microwave

    Theory and Techniques, December 2007 pp 2614 - 2624.

    11. P.G. Huray, O. Oluwafemi, J. Loyer, E. Bogatin, and X. Ye, “Impact of Copper Surface Texture on Loss: A model that works,” DesignCon2010

    12. M Griesi, P.G. Huray, O. Oluwafemi, S. H. Hall, J. Fatcheric, “Electrodeposited copper foil surface characterization for accurate conductor loss modeling,”

    DesignCon2015

    13. L. Simonovich, “Practical method for modeling conductor surface roughness using close packing of equal spheres,” DesignCon2015

    14. A. Deutsch, A.; Huber, G.V. Kopcsay, B. J. Rubin, R. Hemedinger, D. Carey, W. Becker, T Winkel, & B. Chamberlin, “Accuracy of dielectric constant

    measurement using the full sheet resonance technique,” p. 311, ., IEEE

    Symposium on Electrical Performance of Electronic Packaging, 2002

    15. A. Deutsch, C. S. Surovic, R. S. Krabbenhoft, G. V. Kopcsay, B. J. Chamberlin, “Prediction of losses caused by roughness of metallization in printed circuit

    boards,” IEEE Transactions on Advanced Packaging, v. 30, n. 2, May 2007, pp

    279-287.

  • 21

    16. L. Ritchey, J. Zasio, R. Pangier, G. Partida, “High speed signal patch losses related to PCB laminate type and copper roughness,” DesignCon2013

    17. A.F. Horn III, J. W. Reynolds, P. A. LaFrance, J. C. Rautio, “ Effect of conductor profile on the insertion loss, phase constant, and dispersion of thin high frequency

    transmission lines,” DesignCon2010

    18. A. F. Horn, III, J. W. Reynolds, J. C. Rautio, “Conductor profile effects on the propagation constant of microstrip transmission line,” Microwave Symposium

    Digest (MTT), 2010 IEEE MTT-S International, pp 868-871

    19. Y. Shlepnev, C. Nwachukwu, “Roughness characterization for interconnect analysis,” 2011 IEEE Intern. Symposium on Electromagnetic Compatibility, pp

    518-523

    20. M. Y. Koledintseva, O. Y. Kashurkin, T. Vincent, S. Hinaga, “Effective roughness dielectric to represent copper foil roughness in printed circuit boards,”

    DesignCon2015

    21. J. Loyer, “Copper Roughness Electromatgnetics 101,” Printer Circuit Design and Fabrication, November 2015.

    Acknowledgement

    The authors thank Dr. Kristi Pance of the Rogers Innovation Center for many helpful

    insights and Ms. Amie Tworzydlo for the skillfully executed cross-sectional photo of

    figure 3.